id sid tid token lemma pos bracis-28398 1 1 a a DET bracis-28398 1 2 framework framework NOUN bracis-28398 1 3 for for ADP bracis-28398 1 4     SPACE bracis-28398 1 5 characterizing characterize VERB bracis-28398 1 6 what what PRON bracis-28398 1 7 makes make VERB bracis-28398 1 8 an an DET bracis-28398 1 9     SPACE bracis-28398 1 10 instance instance NOUN bracis-28398 1 11 hard hard ADV bracis-28398 1 12 to to PART bracis-28398 1 13     SPACE bracis-28398 1 14 classify classify VERB bracis-28398 1 15 | | ADV bracis-28398 1 16 springer springer NOUN bracis-28398 1 17 nature nature NOUN bracis-28398 1 18 link link PROPN bracis-28398 1 19 ( ( PUNCT bracis-28398 1 20 formerly formerly ADV bracis-28398 1 21 springerlink springerlink NOUN bracis-28398 1 22 ) ) PUNCT bracis-28398 1 23 skip skip VERB bracis-28398 1 24 to to ADP bracis-28398 1 25 main main ADJ bracis-28398 1 26 content content NOUN bracis-28398 1 27 advertisement advertisement NOUN bracis-28398 1 28 log log NOUN bracis-28398 1 29 in in ADP bracis-28398 1 30 menu menu NOUN bracis-28398 1 31 find find VERB bracis-28398 1 32 a a DET bracis-28398 1 33 journal journal NOUN bracis-28398 1 34 publish publish VERB bracis-28398 1 35 with with ADP bracis-28398 1 36 us we PRON bracis-28398 1 37 track track VERB bracis-28398 1 38 your your PRON bracis-28398 1 39 research research NOUN bracis-28398 1 40 search search NOUN bracis-28398 1 41 cart cart NOUN bracis-28398 1 42 home home NOUN bracis-28398 1 43 intelligent intelligent ADJ bracis-28398 1 44 systems system NOUN bracis-28398 1 45 conference conference NOUN bracis-28398 1 46 paper paper NOUN bracis-28398 1 47 a a DET bracis-28398 1 48 framework framework NOUN bracis-28398 1 49 for for ADP bracis-28398 1 50     SPACE bracis-28398 1 51 characterizing characterize VERB bracis-28398 1 52 what what PRON bracis-28398 1 53 makes make VERB bracis-28398 1 54 an an DET bracis-28398 1 55     SPACE bracis-28398 1 56 instance instance NOUN bracis-28398 1 57 hard hard ADV bracis-28398 1 58 to to PART bracis-28398 1 59     SPACE bracis-28398 1 60 classify classify VERB bracis-28398 1 61 conference conference NOUN bracis-28398 1 62 paper paper NOUN bracis-28398 1 63 first first ADV bracis-28398 1 64 online online ADV bracis-28398 1 65 : : PUNCT bracis-28398 1 66 12 12 NUM bracis-28398 1 67 october october NOUN bracis-28398 1 68 2023 2023 NUM bracis-28398 1 69 pp pp PROPN bracis-28398 1 70 353–367 353–367 PROPN bracis-28398 1 71 cite cite VERB bracis-28398 1 72 this this DET bracis-28398 1 73 conference conference NOUN bracis-28398 1 74 paper paper NOUN bracis-28398 1 75 access access NOUN bracis-28398 1 76 provided provide VERB bracis-28398 1 77 by by ADP bracis-28398 1 78 university university PROPN bracis-28398 1 79 of of ADP bracis-28398 1 80 notre notre PROPN bracis-28398 1 81 dame dame PROPN bracis-28398 1 82 hesburgh hesburgh PROPN bracis-28398 1 83 library library PROPN bracis-28398 1 84 download download PROPN bracis-28398 1 85 book book NOUN bracis-28398 1 86 pdf pdf PROPN bracis-28398 1 87 download download NOUN bracis-28398 1 88 book book NOUN bracis-28398 1 89 epub epub PROPN bracis-28398 1 90 intelligent intelligent ADJ bracis-28398 1 91 systems system NOUN bracis-28398 1 92 ( ( PUNCT bracis-28398 1 93 bracis bracis NOUN bracis-28398 1 94 2023 2023 NUM bracis-28398 1 95 ) ) PUNCT bracis-28398 1 96 a a DET bracis-28398 1 97 framework framework NOUN bracis-28398 1 98 for for ADP bracis-28398 1 99     SPACE bracis-28398 1 100 characterizing characterize VERB bracis-28398 1 101 what what PRON bracis-28398 1 102 makes make VERB bracis-28398 1 103 an an DET bracis-28398 1 104     SPACE bracis-28398 1 105 instance instance NOUN bracis-28398 1 106 hard hard ADV bracis-28398 1 107 to to PART bracis-28398 1 108     SPACE bracis-28398 1 109 classify classify VERB bracis-28398 1 110 download download NOUN bracis-28398 1 111 book book NOUN bracis-28398 1 112 pdf pdf PROPN bracis-28398 1 113 download download NOUN bracis-28398 1 114 book book PROPN bracis-28398 1 115 epub epub PROPN bracis-28398 1 116 maria maria PROPN bracis-28398 1 117 gabriela gabriela PROPN bracis-28398 1 118 valeriano valeriano PROPN bracis-28398 1 119     SPACE bracis-28398 1 120 orcid orcid NOUN bracis-28398 1 121 : : PUNCT bracis-28398 1 122 orcid.org/0000-0002-3631-156x9,10 orcid.org/0000-0002-3631-156x9,10 NOUN bracis-28398 1 123 , , PUNCT bracis-28398 1 124 pedro pedro PROPN bracis-28398 1 125 yuri yuri PROPN bracis-28398 1 126 arbs arbs PROPN bracis-28398 1 127 paiva paiva PROPN bracis-28398 1 128     SPACE bracis-28398 1 129 orcid orcid PROPN bracis-28398 1 130 : : PUNCT bracis-28398 2 1 orcid.org/0000-0001-9768-75639 orcid.org/0000-0001-9768-75639 PROPN bracis-28398 2 2 , , PUNCT bracis-28398 2 3 carlos carlos PROPN bracis-28398 2 4 roberto roberto PROPN bracis-28398 2 5 veiga veiga PROPN bracis-28398 2 6 kiffer kiffer PROPN bracis-28398 2 7     SPACE bracis-28398 2 8 orcid orcid PROPN bracis-28398 2 9 : : PUNCT bracis-28398 2 10 orcid.org/0000-0003-1122-069310 orcid.org/0000-0003-1122-069310 PROPN bracis-28398 2 11 & & CCONJ bracis-28398 2 12 … … PUNCT bracis-28398 2 13 ana ana PROPN bracis-28398 2 14 carolina carolina PROPN bracis-28398 2 15 lorena lorena PROPN bracis-28398 2 16     SPACE bracis-28398 2 17 orcid orcid PROPN bracis-28398 2 18 : : PUNCT bracis-28398 2 19 orcid.org/0000-0002-6140-571x9 orcid.org/0000-0002-6140-571x9 NUM bracis-28398 2 20     SPACE bracis-28398 2 21 show show VERB bracis-28398 2 22 authors author NOUN bracis-28398 2 23 part part NOUN bracis-28398 2 24 of of ADP bracis-28398 2 25 the the DET bracis-28398 2 26 book book NOUN bracis-28398 2 27 series series NOUN bracis-28398 2 28 : : PUNCT bracis-28398 2 29 lecture lecture NOUN bracis-28398 2 30 notes note NOUN bracis-28398 2 31 in in ADP bracis-28398 2 32 computer computer NOUN bracis-28398 2 33 science science NOUN bracis-28398 2 34 ( ( PUNCT bracis-28398 2 35 ( ( PUNCT bracis-28398 2 36 lnai lnai ADJ bracis-28398 2 37 , , PUNCT bracis-28398 2 38 volume volume NOUN bracis-28398 2 39 14196 14196 NUM bracis-28398 2 40 ) ) PUNCT bracis-28398 2 41 ) ) PUNCT bracis-28398 2 42 included include VERB bracis-28398 2 43 in in ADP bracis-28398 2 44 the the DET bracis-28398 2 45 following follow VERB bracis-28398 2 46 conference conference NOUN bracis-28398 2 47 series series NOUN bracis-28398 2 48 : : PUNCT bracis-28398 2 49 brazilian brazilian ADJ bracis-28398 2 50 conference conference NOUN bracis-28398 2 51 on on ADP bracis-28398 2 52 intelligent intelligent ADJ bracis-28398 2 53 systems system NOUN bracis-28398 2 54 586 586 NUM bracis-28398 2 55 accesses access NOUN bracis-28398 2 56 1 1 NUM bracis-28398 2 57 citation citation NOUN bracis-28398 2 58 abstract abstract NOUN bracis-28398 2 59 the the DET bracis-28398 2 60 health health NOUN bracis-28398 2 61 domain domain NOUN bracis-28398 2 62 has have AUX bracis-28398 2 63 been be AUX bracis-28398 2 64 largely largely ADV bracis-28398 2 65 benefited benefit VERB bracis-28398 2 66 by by ADP bracis-28398 2 67 machine machine NOUN bracis-28398 2 68 learning learn VERB bracis-28398 2 69 solutions solution NOUN bracis-28398 2 70 , , PUNCT bracis-28398 2 71 which which PRON bracis-28398 2 72 can can AUX bracis-28398 2 73 be be AUX bracis-28398 2 74 used use VERB bracis-28398 2 75 for for ADP bracis-28398 2 76 building build VERB bracis-28398 2 77 predictive predictive ADJ bracis-28398 2 78 models model NOUN bracis-28398 2 79 to to PART bracis-28398 2 80 support support VERB bracis-28398 2 81 medical medical ADJ bracis-28398 2 82 decisions decision NOUN bracis-28398 2 83 . . PUNCT bracis-28398 3 1 but but CCONJ bracis-28398 3 2 , , PUNCT bracis-28398 3 3 for for ADP bracis-28398 3 4 increasing increase VERB bracis-28398 3 5 the the DET bracis-28398 3 6 reliability reliability NOUN bracis-28398 3 7 of of ADP bracis-28398 3 8 these these DET bracis-28398 3 9 systems system NOUN bracis-28398 3 10 , , PUNCT bracis-28398 3 11 it it PRON bracis-28398 3 12 is be AUX bracis-28398 3 13 important important ADJ bracis-28398 3 14 to to PART bracis-28398 3 15 understand understand VERB bracis-28398 3 16 when when SCONJ bracis-28398 3 17 the the DET bracis-28398 3 18 models model NOUN bracis-28398 3 19 are be AUX bracis-28398 3 20 prone prone ADJ bracis-28398 3 21 to to ADP bracis-28398 3 22 failures failure NOUN bracis-28398 3 23 . . PUNCT bracis-28398 4 1 in in ADP bracis-28398 4 2 this this DET bracis-28398 4 3 paper paper NOUN bracis-28398 4 4 , , PUNCT bracis-28398 4 5 we we PRON bracis-28398 4 6 investigate investigate VERB bracis-28398 4 7 what what PRON bracis-28398 4 8 can can AUX bracis-28398 4 9 we we PRON bracis-28398 4 10 learn learn VERB bracis-28398 4 11 from from ADP bracis-28398 4 12 the the DET bracis-28398 4 13 instances instance NOUN bracis-28398 4 14 of of ADP bracis-28398 4 15 a a DET bracis-28398 4 16 dataset dataset NOUN bracis-28398 4 17 which which PRON bracis-28398 4 18 are be AUX bracis-28398 4 19 hard hard ADJ bracis-28398 4 20 to to PART bracis-28398 4 21 classify classify VERB bracis-28398 4 22 by by ADP bracis-28398 4 23 machine machine NOUN bracis-28398 4 24 learning learning NOUN bracis-28398 4 25 models model NOUN bracis-28398 4 26 . . PUNCT bracis-28398 5 1 different different ADJ bracis-28398 5 2 reasons reason NOUN bracis-28398 5 3 may may AUX bracis-28398 5 4 explain explain VERB bracis-28398 5 5 why why SCONJ bracis-28398 5 6 one one NUM bracis-28398 5 7 or or CCONJ bracis-28398 5 8 a a DET bracis-28398 5 9 set set NOUN bracis-28398 5 10 of of ADP bracis-28398 5 11 instances instance NOUN bracis-28398 5 12 are be AUX bracis-28398 5 13 misclassified misclassifie VERB bracis-28398 5 14 , , PUNCT bracis-28398 5 15 despite despite SCONJ bracis-28398 5 16 the the DET bracis-28398 5 17 predictive predictive ADJ bracis-28398 5 18 model model NOUN bracis-28398 5 19 used use VERB bracis-28398 5 20 . . PUNCT bracis-28398 6 1 they they PRON bracis-28398 6 2 can can AUX bracis-28398 6 3 be be AUX bracis-28398 6 4 either either CCONJ bracis-28398 6 5 noisy noisy ADJ bracis-28398 6 6 , , PUNCT bracis-28398 6 7 anomalous anomalous ADJ bracis-28398 6 8 or or CCONJ bracis-28398 6 9 placed place VERB bracis-28398 6 10 in in ADP bracis-28398 6 11 overlapping overlap VERB bracis-28398 6 12 regions region NOUN bracis-28398 6 13 , , PUNCT bracis-28398 6 14 to to PART bracis-28398 6 15 name name VERB bracis-28398 6 16 a a DET bracis-28398 6 17 few few ADJ bracis-28398 6 18 . . PUNCT bracis-28398 7 1 our our PRON bracis-28398 7 2 framework framework NOUN bracis-28398 7 3 works work VERB bracis-28398 7 4 at at ADP bracis-28398 7 5 two two NUM bracis-28398 7 6 levels level NOUN bracis-28398 7 7 : : PUNCT bracis-28398 7 8 the the DET bracis-28398 7 9 original original ADJ bracis-28398 7 10 base base NOUN bracis-28398 7 11 dataset dataset NOUN bracis-28398 7 12 and and CCONJ bracis-28398 7 13 a a DET bracis-28398 7 14 meta meta ADJ bracis-28398 7 15 - - PUNCT bracis-28398 7 16 dataset dataset NOUN bracis-28398 7 17 built build VERB bracis-28398 7 18 to to PART bracis-28398 7 19 reflect reflect VERB bracis-28398 7 20 the the DET bracis-28398 7 21 hardness hardness ADJ bracis-28398 7 22 level level NOUN bracis-28398 7 23 of of ADP bracis-28398 7 24 the the DET bracis-28398 7 25 instances instance NOUN bracis-28398 7 26 . . PUNCT bracis-28398 8 1 a a DET bracis-28398 8 2 two two NUM bracis-28398 8 3 - - PUNCT bracis-28398 8 4 dimensional dimensional ADJ bracis-28398 8 5 hardness hardness NOUN bracis-28398 8 6 embedding embedding NOUN bracis-28398 8 7 is be AUX bracis-28398 8 8 assembled assemble VERB bracis-28398 8 9 , , PUNCT bracis-28398 8 10 which which PRON bracis-28398 8 11 can can AUX bracis-28398 8 12 be be AUX bracis-28398 8 13 visually visually ADV bracis-28398 8 14 inspected inspect VERB bracis-28398 8 15 to to PART bracis-28398 8 16 determine determine VERB bracis-28398 8 17 sets set NOUN bracis-28398 8 18 of of ADP bracis-28398 8 19 instances instance NOUN bracis-28398 8 20 to to PART bracis-28398 8 21 scrutinize scrutinize VERB bracis-28398 8 22 better well ADV bracis-28398 8 23 . . PUNCT bracis-28398 9 1 we we PRON bracis-28398 9 2 show show VERB bracis-28398 9 3 some some DET bracis-28398 9 4 analysis analysis NOUN bracis-28398 9 5 that that PRON bracis-28398 9 6 can can AUX bracis-28398 9 7 be be AUX bracis-28398 9 8 undertaken undertake VERB bracis-28398 9 9 in in ADP bracis-28398 9 10 this this DET bracis-28398 9 11 hardness hardness ADJ bracis-28398 9 12 space space NOUN bracis-28398 9 13 that that PRON bracis-28398 9 14 allow allow VERB bracis-28398 9 15 to to PART bracis-28398 9 16 characterize characterize VERB bracis-28398 9 17 why why SCONJ bracis-28398 9 18 some some PRON bracis-28398 9 19 of of ADP bracis-28398 9 20 the the DET bracis-28398 9 21 instances instance NOUN bracis-28398 9 22 are be AUX bracis-28398 9 23 hard hard ADJ bracis-28398 9 24 to to PART bracis-28398 9 25 classify classify VERB bracis-28398 9 26 , , PUNCT bracis-28398 9 27 with with ADP bracis-28398 9 28 case case NOUN bracis-28398 9 29 studies study NOUN bracis-28398 9 30 on on ADP bracis-28398 9 31 health health NOUN bracis-28398 9 32 datasets dataset NOUN bracis-28398 9 33 . . PUNCT bracis-28398 10 1 access access NOUN bracis-28398 10 2 provided provide VERB bracis-28398 10 3 by by ADP bracis-28398 10 4 university university PROPN bracis-28398 10 5 of of ADP bracis-28398 10 6 notre notre PROPN bracis-28398 10 7 dame dame PROPN bracis-28398 10 8 hesburgh hesburgh PROPN bracis-28398 10 9 library library PROPN bracis-28398 10 10 . . PUNCT bracis-28398 11 1 download download PROPN bracis-28398 11 2 conference conference NOUN bracis-28398 11 3 paper paper NOUN bracis-28398 11 4 pdf pdf NOUN bracis-28398 11 5 similar similar ADJ bracis-28398 11 6 content content NOUN bracis-28398 11 7 being be AUX bracis-28398 11 8 viewed view VERB bracis-28398 11 9 by by ADP bracis-28398 11 10 others other NOUN bracis-28398 11 11 comparative comparative ADJ bracis-28398 11 12 analysis analysis NOUN bracis-28398 11 13 of of ADP bracis-28398 11 14 ensemble ensemble ADJ bracis-28398 11 15 classifier classifier NOUN bracis-28398 11 16 and and CCONJ bracis-28398 11 17 single single ADJ bracis-28398 11 18 base base NOUN bracis-28398 11 19 classifier classifier NOUN bracis-28398 11 20 in in ADP bracis-28398 11 21 medical medical ADJ bracis-28398 11 22 disease disease NOUN bracis-28398 11 23 diagnosis diagnosis NOUN bracis-28398 11 24 chapter chapter NOUN bracis-28398 11 25 © © PROPN bracis-28398 11 26 2020 2020 NUM bracis-28398 11 27 a a DET bracis-28398 11 28 meta meta ADJ bracis-28398 11 29 - - PUNCT bracis-28398 11 30 heuristic heuristic ADJ bracis-28398 11 31 approach approach NOUN bracis-28398 11 32 to to PART bracis-28398 11 33 estimate estimate VERB bracis-28398 11 34 and and CCONJ bracis-28398 11 35 explain explain VERB bracis-28398 11 36 classifier classifier ADJ bracis-28398 11 37 uncertainty uncertainty NOUN bracis-28398 11 38 article article NOUN bracis-28398 11 39 open open VERB bracis-28398 11 40 access access NOUN bracis-28398 11 41 14 14 NUM bracis-28398 11 42 january january PROPN bracis-28398 11 43 2025 2025 NUM bracis-28398 11 44 machine machine NOUN bracis-28398 11 45 learning learning NOUN bracis-28398 11 46 in in ADP bracis-28398 11 47 practice practice NOUN bracis-28398 11 48 — — PUNCT bracis-28398 11 49 clinical clinical ADJ bracis-28398 11 50 decision decision NOUN bracis-28398 11 51 support support NOUN bracis-28398 11 52 , , PUNCT bracis-28398 11 53 risk risk NOUN bracis-28398 11 54 prediction prediction NOUN bracis-28398 11 55 , , PUNCT bracis-28398 11 56 diagnosis diagnosis NOUN bracis-28398 11 57 chapter chapter NOUN bracis-28398 11 58 © © PROPN bracis-28398 11 59 2023 2023 NUM bracis-28398 11 60 explore explore VERB bracis-28398 11 61 related relate VERB bracis-28398 11 62 subjects subject NOUN bracis-28398 11 63 discover discover VERB bracis-28398 11 64 the the DET bracis-28398 11 65 latest late ADJ bracis-28398 11 66 articles article NOUN bracis-28398 11 67 , , PUNCT bracis-28398 11 68 books book NOUN bracis-28398 11 69 and and CCONJ bracis-28398 11 70 news news NOUN bracis-28398 11 71 in in ADP bracis-28398 11 72 related related ADJ bracis-28398 11 73 subjects subject NOUN bracis-28398 11 74 , , PUNCT bracis-28398 11 75 suggested suggest VERB bracis-28398 11 76 using use VERB bracis-28398 11 77 machine machine NOUN bracis-28398 11 78 learning learning NOUN bracis-28398 11 79 . . PUNCT bracis-28398 12 1 categorization categorization NOUN bracis-28398 12 2 complexity complexity NOUN bracis-28398 12 3 data datum NOUN bracis-28398 12 4 mining mining NOUN bracis-28398 12 5 learning learn VERB bracis-28398 12 6 algorithms algorithm NOUN bracis-28398 12 7 machine machine NOUN bracis-28398 12 8 learning learn VERB bracis-28398 12 9 statistical statistical ADJ bracis-28398 12 10 learning learn VERB bracis-28398 12 11 1 1 NUM bracis-28398 12 12 introduction introduction NOUN bracis-28398 12 13 machine machine NOUN bracis-28398 12 14 learning learning NOUN bracis-28398 12 15 ( ( PUNCT bracis-28398 12 16 ml ml NOUN bracis-28398 12 17 ) ) PUNCT bracis-28398 12 18 models model NOUN bracis-28398 12 19 are be AUX bracis-28398 12 20 regarded regard VERB bracis-28398 12 21 as as ADP bracis-28398 12 22 promising promise VERB bracis-28398 12 23 solutions solution NOUN bracis-28398 12 24 for for ADP bracis-28398 12 25 revolutionizing revolutionize VERB bracis-28398 12 26 healthcare healthcare PROPN bracis-28398 12 27     SPACE bracis-28398 13 1 [ [ X bracis-28398 13 2 1 1 NUM bracis-28398 13 3 ] ] PUNCT bracis-28398 13 4 . . PUNCT bracis-28398 14 1 these these DET bracis-28398 14 2 data data NOUN bracis-28398 14 3 - - PUNCT bracis-28398 14 4 driven drive VERB bracis-28398 14 5 techniques technique NOUN bracis-28398 14 6 may may AUX bracis-28398 14 7 leverage leverage VERB bracis-28398 14 8 knowledge knowledge NOUN bracis-28398 14 9 from from ADP bracis-28398 14 10 the the DET bracis-28398 14 11 large large ADJ bracis-28398 14 12 volumes volume NOUN bracis-28398 14 13 of of ADP bracis-28398 14 14 data datum NOUN bracis-28398 14 15 continuously continuously ADV bracis-28398 14 16 gathered gather VERB bracis-28398 14 17 by by ADP bracis-28398 14 18 health health NOUN bracis-28398 14 19 systems system NOUN bracis-28398 14 20 and and CCONJ bracis-28398 14 21 agents agent NOUN bracis-28398 14 22     SPACE bracis-28398 15 1 [ [ X bracis-28398 15 2 2 2 NUM bracis-28398 15 3 ] ] PUNCT bracis-28398 15 4 . . PUNCT bracis-28398 16 1 as as ADP bracis-28398 16 2 a a DET bracis-28398 16 3 recent recent ADJ bracis-28398 16 4 example example NOUN bracis-28398 16 5 , , PUNCT bracis-28398 16 6 different different ADJ bracis-28398 16 7 predictive predictive ADJ bracis-28398 16 8 models model NOUN bracis-28398 16 9 were be AUX bracis-28398 16 10 built build VERB bracis-28398 16 11 during during ADP bracis-28398 16 12 the the DET bracis-28398 16 13 covid-19 covid-19 PROPN bracis-28398 16 14 pandemic pandemic NOUN bracis-28398 16 15 to to PART bracis-28398 16 16 assist assist VERB bracis-28398 16 17 diagnosis diagnosis NOUN bracis-28398 16 18 and and CCONJ bracis-28398 16 19 prognosis prognosis NOUN bracis-28398 16 20 of of ADP bracis-28398 16 21 patients patient NOUN bracis-28398 16 22 using use VERB bracis-28398 16 23 hospital hospital NOUN bracis-28398 16 24 data datum NOUN bracis-28398 16 25 [ [ X bracis-28398 16 26 3 3 NUM bracis-28398 16 27 , , PUNCT bracis-28398 16 28 4 4 NUM bracis-28398 16 29 ] ] PUNCT bracis-28398 16 30 . . PUNCT bracis-28398 17 1 nonetheless nonetheless ADV bracis-28398 17 2 , , PUNCT bracis-28398 17 3 there there PRON bracis-28398 17 4 are be VERB bracis-28398 17 5 issues issue NOUN bracis-28398 17 6 still still ADV bracis-28398 17 7 preventing prevent VERB bracis-28398 17 8 the the DET bracis-28398 17 9 widespread widespread ADJ bracis-28398 17 10 usage usage NOUN bracis-28398 17 11 of of ADP bracis-28398 17 12 ml ml NOUN bracis-28398 17 13 predictive predictive ADJ bracis-28398 17 14 models model NOUN bracis-28398 17 15 in in ADP bracis-28398 17 16 health health NOUN bracis-28398 17 17 decision decision NOUN bracis-28398 17 18 making making NOUN bracis-28398 17 19 and and CCONJ bracis-28398 17 20 planning planning NOUN bracis-28398 17 21 . . PUNCT bracis-28398 18 1 some some DET bracis-28398 18 2 concerns concern NOUN bracis-28398 18 3 are be AUX bracis-28398 18 4 the the DET bracis-28398 18 5 risk risk NOUN bracis-28398 18 6 of of ADP bracis-28398 18 7 bias bias NOUN bracis-28398 18 8 and and CCONJ bracis-28398 18 9 inappropriate inappropriate ADJ bracis-28398 18 10 or or CCONJ bracis-28398 18 11 incomplete incomplete ADJ bracis-28398 18 12 model model NOUN bracis-28398 18 13 performance performance NOUN bracis-28398 18 14 evaluation evaluation NOUN bracis-28398 18 15     SPACE bracis-28398 19 1 [ [ X bracis-28398 19 2 5 5 NUM bracis-28398 19 3 ] ] PUNCT bracis-28398 19 4 . . PUNCT bracis-28398 20 1 given give VERB bracis-28398 20 2 that that SCONJ bracis-28398 20 3 health health NOUN bracis-28398 20 4 databases database NOUN bracis-28398 20 5 are be AUX bracis-28398 20 6 not not PART bracis-28398 20 7 collected collect VERB bracis-28398 20 8 with with ADP bracis-28398 20 9 the the DET bracis-28398 20 10 objective objective NOUN bracis-28398 20 11 of of ADP bracis-28398 20 12 data datum NOUN bracis-28398 20 13 analysis analysis NOUN bracis-28398 20 14 in in ADP bracis-28398 20 15 the the DET bracis-28398 20 16 first first ADJ bracis-28398 20 17 place place NOUN bracis-28398 20 18 , , PUNCT bracis-28398 20 19 the the DET bracis-28398 20 20 data data NOUN bracis-28398 20 21 scientist scientist NOUN bracis-28398 20 22 and and CCONJ bracis-28398 20 23 ml ml ADP bracis-28398 20 24 practitioner practitioner NOUN bracis-28398 20 25 must must AUX bracis-28398 20 26 deal deal VERB bracis-28398 20 27 with with ADP bracis-28398 20 28 these these DET bracis-28398 20 29 issues issue NOUN bracis-28398 20 30 by by ADP bracis-28398 20 31 relying rely VERB bracis-28398 20 32 on on ADP bracis-28398 20 33 their their PRON bracis-28398 20 34 own own ADJ bracis-28398 20 35 experience experience NOUN bracis-28398 20 36 or or CCONJ bracis-28398 20 37 on on ADP bracis-28398 20 38 ad ad NOUN bracis-28398 20 39 - - PUNCT bracis-28398 20 40 hoc hoc X bracis-28398 20 41 procedures procedure NOUN bracis-28398 20 42 , , PUNCT bracis-28398 20 43 which which PRON bracis-28398 20 44 can can AUX bracis-28398 20 45 bias bias VERB bracis-28398 20 46 the the DET bracis-28398 20 47 achieved achieve VERB bracis-28398 20 48 results result NOUN bracis-28398 20 49 . . PUNCT bracis-28398 21 1 data datum NOUN bracis-28398 21 2 - - PUNCT bracis-28398 21 3 centric centric ADJ bracis-28398 21 4 frameworks framework NOUN bracis-28398 21 5 have have AUX bracis-28398 21 6 recently recently ADV bracis-28398 21 7 been be AUX bracis-28398 21 8 proposed propose VERB bracis-28398 21 9 to to PART bracis-28398 21 10 aid aid VERB bracis-28398 21 11 these these DET bracis-28398 21 12 professionals professional NOUN bracis-28398 21 13 in in ADP bracis-28398 21 14 better well ADJ bracis-28398 21 15 assessing assess VERB bracis-28398 21 16 data datum NOUN bracis-28398 21 17 quality quality NOUN bracis-28398 21 18 and and CCONJ bracis-28398 21 19 taking take VERB bracis-28398 21 20 potential potential ADJ bracis-28398 21 21 corrective corrective ADJ bracis-28398 21 22 measures measure NOUN bracis-28398 21 23 , , PUNCT bracis-28398 21 24 which which PRON bracis-28398 21 25 will will AUX bracis-28398 21 26 result result VERB bracis-28398 21 27 in in ADP bracis-28398 21 28 more more ADV bracis-28398 21 29 reliable reliable ADJ bracis-28398 21 30 and and CCONJ bracis-28398 21 31 trustful trustful ADJ bracis-28398 21 32 predictive predictive ADJ bracis-28398 21 33 models model NOUN bracis-28398 21 34     SPACE bracis-28398 22 1 [ [ X bracis-28398 22 2 6 6 NUM bracis-28398 22 3 , , PUNCT bracis-28398 22 4 7 7 NUM bracis-28398 22 5 ] ] PUNCT bracis-28398 22 6 . . PUNCT bracis-28398 23 1 one one NUM bracis-28398 23 2 fruitful fruitful ADJ bracis-28398 23 3 direction direction NOUN bracis-28398 23 4 is be AUX bracis-28398 23 5 to to PART bracis-28398 23 6 assess assess VERB bracis-28398 23 7 and and CCONJ bracis-28398 23 8 monitor monitor VERB bracis-28398 23 9 learning learn VERB bracis-28398 23 10 performance performance NOUN bracis-28398 23 11 for for ADP bracis-28398 23 12 each each DET bracis-28398 23 13 dataset dataset NOUN bracis-28398 23 14 instance instance NOUN bracis-28398 23 15 and and CCONJ bracis-28398 23 16 prevent prevent VERB bracis-28398 23 17 relying rely VERB bracis-28398 23 18 only only ADV bracis-28398 23 19 on on ADP bracis-28398 23 20 averages average NOUN bracis-28398 23 21 over over ADP bracis-28398 23 22 an an DET bracis-28398 23 23 entire entire ADJ bracis-28398 23 24 dataset dataset NOUN bracis-28398 23 25 , , PUNCT bracis-28398 23 26 as as SCONJ bracis-28398 23 27 it it PRON bracis-28398 23 28 is be AUX bracis-28398 23 29 common common ADJ bracis-28398 23 30 practice practice NOUN bracis-28398 23 31     SPACE bracis-28398 24 1 [ [ X bracis-28398 24 2 8 8 NUM bracis-28398 24 3 ] ] PUNCT bracis-28398 24 4 . . PUNCT bracis-28398 25 1 furthermore furthermore ADV bracis-28398 25 2 , , PUNCT bracis-28398 25 3 examining examine VERB bracis-28398 25 4 the the DET bracis-28398 25 5 performance performance NOUN bracis-28398 25 6 at at ADP bracis-28398 25 7 the the DET bracis-28398 25 8 instance instance NOUN bracis-28398 25 9 level level NOUN bracis-28398 25 10 can can AUX bracis-28398 25 11 reduce reduce VERB bracis-28398 25 12 biases bias NOUN bracis-28398 25 13 in in ADP bracis-28398 25 14 the the DET bracis-28398 25 15 evaluation evaluation NOUN bracis-28398 25 16 of of ADP bracis-28398 25 17 ml ml NOUN bracis-28398 25 18 models model NOUN bracis-28398 25 19 , , PUNCT bracis-28398 25 20 as as SCONJ bracis-28398 25 21 specific specific ADJ bracis-28398 25 22 important important ADJ bracis-28398 25 23 groups group NOUN bracis-28398 25 24 of of ADP bracis-28398 25 25 instances instance NOUN bracis-28398 25 26 can can AUX bracis-28398 25 27 be be AUX bracis-28398 25 28 mistaken mistake VERB bracis-28398 25 29 by by ADP bracis-28398 25 30 the the DET bracis-28398 25 31 ml ml PROPN bracis-28398 25 32 models model NOUN bracis-28398 25 33 , , PUNCT bracis-28398 25 34 despite despite SCONJ bracis-28398 25 35 the the DET bracis-28398 25 36 obtainment obtainment NOUN bracis-28398 25 37 of of ADP bracis-28398 25 38 an an DET bracis-28398 25 39 overall overall ADJ bracis-28398 25 40 high high ADJ bracis-28398 25 41 average average ADJ bracis-28398 25 42 predictive predictive ADJ bracis-28398 25 43 performance performance NOUN bracis-28398 25 44 . . PUNCT bracis-28398 26 1 this this PRON bracis-28398 26 2 makes make VERB bracis-28398 26 3 it it PRON bracis-28398 26 4 important important ADJ bracis-28398 26 5 to to PART bracis-28398 26 6 know know VERB bracis-28398 26 7 which which DET bracis-28398 26 8 instances instance NOUN bracis-28398 26 9 are be AUX bracis-28398 26 10 systematically systematically ADV bracis-28398 26 11 misclassified misclassifie VERB bracis-28398 26 12 by by ADP bracis-28398 26 13 ml ml NOUN bracis-28398 26 14 models model NOUN bracis-28398 26 15 and and CCONJ bracis-28398 26 16 why why SCONJ bracis-28398 26 17 they they PRON bracis-28398 26 18 are be AUX bracis-28398 26 19 so so ADV bracis-28398 26 20 , , PUNCT bracis-28398 26 21 a a DET bracis-28398 26 22 concern concern NOUN bracis-28398 26 23 that that PRON bracis-28398 26 24 has have AUX bracis-28398 26 25 led lead VERB bracis-28398 26 26 to to ADP bracis-28398 26 27 the the DET bracis-28398 26 28 recent recent ADJ bracis-28398 26 29 literature literature NOUN bracis-28398 26 30 of of ADP bracis-28398 26 31 instance instance NOUN bracis-28398 26 32 hardness hardness NOUN bracis-28398 26 33 analysis analysis NOUN bracis-28398 26 34     SPACE bracis-28398 27 1 [ [ X bracis-28398 27 2 9 9 NUM bracis-28398 27 3 ] ] PUNCT bracis-28398 27 4 . . PUNCT bracis-28398 28 1 according accord VERB bracis-28398 28 2 to to ADP bracis-28398 28 3 smith smith PROPN bracis-28398 28 4 et et PROPN bracis-28398 28 5 al al PROPN bracis-28398 28 6 . . PUNCT bracis-28398 28 7     SPACE bracis-28398 29 1 [ [ X bracis-28398 29 2 9 9 NUM bracis-28398 29 3 ] ] PUNCT bracis-28398 29 4 , , PUNCT bracis-28398 29 5 instance instance NOUN bracis-28398 29 6 hardness hardness NOUN bracis-28398 29 7 can can AUX bracis-28398 29 8 be be AUX bracis-28398 29 9 measured measure VERB bracis-28398 29 10 as as ADP bracis-28398 29 11 the the DET bracis-28398 29 12 average average ADJ bracis-28398 29 13 misclassification misclassification NOUN bracis-28398 29 14 error error NOUN bracis-28398 29 15 of of ADP bracis-28398 29 16 a a DET bracis-28398 29 17 pool pool NOUN bracis-28398 29 18 of of ADP bracis-28398 29 19 diverse diverse ADJ bracis-28398 29 20 classifiers classifier NOUN bracis-28398 29 21 when when SCONJ bracis-28398 29 22 predicting predict VERB bracis-28398 29 23 the the DET bracis-28398 29 24 label label NOUN bracis-28398 29 25 of of ADP bracis-28398 29 26 such such ADJ bracis-28398 29 27 instance instance NOUN bracis-28398 29 28 . . PUNCT bracis-28398 30 1 this this DET bracis-28398 30 2 metric metric NOUN bracis-28398 30 3 allows allow VERB bracis-28398 30 4 identifying identify VERB bracis-28398 30 5 instances instance NOUN bracis-28398 30 6 in in ADP bracis-28398 30 7 a a DET bracis-28398 30 8 dataset dataset NOUN bracis-28398 30 9 that that PRON bracis-28398 30 10 are be AUX bracis-28398 30 11 inherently inherently ADV bracis-28398 30 12 difficult difficult ADJ bracis-28398 30 13 to to PART bracis-28398 30 14 have have VERB bracis-28398 30 15 their their PRON bracis-28398 30 16 label label NOUN bracis-28398 30 17 predicted predict VERB bracis-28398 30 18 , , PUNCT bracis-28398 30 19 despite despite SCONJ bracis-28398 30 20 of of ADP bracis-28398 30 21 the the DET bracis-28398 30 22 classification classification NOUN bracis-28398 30 23 technique technique NOUN bracis-28398 30 24 used use VERB bracis-28398 30 25 . . PUNCT bracis-28398 31 1 however however ADV bracis-28398 31 2 , , PUNCT bracis-28398 31 3 going go VERB bracis-28398 31 4 beyond beyond ADV bracis-28398 31 5 , , PUNCT bracis-28398 31 6 the the DET bracis-28398 31 7 literature literature NOUN bracis-28398 31 8 also also ADV bracis-28398 31 9 presents present VERB bracis-28398 31 10 hardness hardness ADJ bracis-28398 31 11 meta meta X bracis-28398 31 12 - - PUNCT bracis-28398 31 13 features feature NOUN bracis-28398 31 14 which which PRON bracis-28398 31 15 prospect prospect VERB bracis-28398 31 16 possible possible ADJ bracis-28398 31 17 reasons reason NOUN bracis-28398 31 18 why why SCONJ bracis-28398 31 19 the the DET bracis-28398 31 20 instance instance NOUN bracis-28398 31 21 is be AUX bracis-28398 31 22 hard hard ADJ bracis-28398 31 23 to to PART bracis-28398 31 24 classify classify VERB bracis-28398 31 25     SPACE bracis-28398 32 1 [ [ X bracis-28398 32 2 9 9 NUM bracis-28398 32 3 , , PUNCT bracis-28398 32 4 10 10 NUM bracis-28398 32 5 ] ] PUNCT bracis-28398 32 6 . . PUNCT bracis-28398 33 1 recently recently ADV bracis-28398 33 2 , , PUNCT bracis-28398 33 3 a a DET bracis-28398 33 4 framework framework NOUN bracis-28398 33 5 for for ADP bracis-28398 33 6 relating relate VERB bracis-28398 33 7 the the DET bracis-28398 33 8 classification classification NOUN bracis-28398 33 9 performance performance NOUN bracis-28398 33 10 of of ADP bracis-28398 33 11 classifiers classifier NOUN bracis-28398 33 12 of of ADP bracis-28398 33 13 distinct distinct ADJ bracis-28398 33 14 biases bias NOUN bracis-28398 33 15 and and CCONJ bracis-28398 33 16 the the DET bracis-28398 33 17 values value NOUN bracis-28398 33 18 of of ADP bracis-28398 33 19 different different ADJ bracis-28398 33 20 hardness hardness NOUN bracis-28398 33 21 meta meta X bracis-28398 33 22 - - PUNCT bracis-28398 33 23 features feature NOUN bracis-28398 33 24 was be AUX bracis-28398 33 25 framed frame VERB bracis-28398 33 26     SPACE bracis-28398 33 27 [ [ X bracis-28398 33 28 8 8 NUM bracis-28398 33 29 , , PUNCT bracis-28398 33 30 11 11 NUM bracis-28398 33 31 ] ] PUNCT bracis-28398 33 32 . . PUNCT bracis-28398 34 1 the the DET bracis-28398 34 2 relationship relationship NOUN bracis-28398 34 3 of of ADP bracis-28398 34 4 this this DET bracis-28398 34 5 information information NOUN bracis-28398 34 6 is be AUX bracis-28398 34 7 used use VERB bracis-28398 34 8 to to PART bracis-28398 34 9 produce produce VERB bracis-28398 34 10 a a DET bracis-28398 34 11 2 2 NUM bracis-28398 34 12 - - PUNCT bracis-28398 34 13 d d NOUN bracis-28398 34 14 hardness hardness NOUN bracis-28398 34 15 embedding embed VERB bracis-28398 34 16 where where SCONJ bracis-28398 34 17 the the DET bracis-28398 34 18 instances instance NOUN bracis-28398 34 19 are be AUX bracis-28398 34 20 linearly linearly ADV bracis-28398 34 21 distributed distribute VERB bracis-28398 34 22 according accord VERB bracis-28398 34 23 to to ADP bracis-28398 34 24 their their PRON bracis-28398 34 25 hardness hardness ADJ bracis-28398 34 26 level level NOUN bracis-28398 34 27 . . PUNCT bracis-28398 35 1 here here ADV bracis-28398 35 2 this this DET bracis-28398 35 3 framework framework NOUN bracis-28398 35 4 is be AUX bracis-28398 35 5 extended extend VERB bracis-28398 35 6 for for ADP bracis-28398 35 7 gaining gain VERB bracis-28398 35 8 data data NOUN bracis-28398 35 9 - - PUNCT bracis-28398 35 10 centric centric ADJ bracis-28398 35 11 insights insight NOUN bracis-28398 35 12 by by ADP bracis-28398 35 13 taking take VERB bracis-28398 35 14 advantage advantage NOUN bracis-28398 35 15 of of ADP bracis-28398 35 16 the the DET bracis-28398 35 17 organization organization NOUN bracis-28398 35 18 of of ADP bracis-28398 35 19 the the DET bracis-28398 35 20 dataset dataset NOUN bracis-28398 35 21 observations observation NOUN bracis-28398 35 22 in in ADP bracis-28398 35 23 the the DET bracis-28398 35 24 hardness hardness NOUN bracis-28398 35 25 embedding embedding NOUN bracis-28398 35 26 . . PUNCT bracis-28398 36 1 these these DET bracis-28398 36 2 analyzes analyze NOUN bracis-28398 36 3 take take VERB bracis-28398 36 4 place place NOUN bracis-28398 36 5 at at ADP bracis-28398 36 6 two two NUM bracis-28398 36 7 levels level NOUN bracis-28398 36 8 : : PUNCT bracis-28398 36 9 the the DET bracis-28398 36 10 original original ADJ bracis-28398 36 11 input input NOUN bracis-28398 36 12 features feature NOUN bracis-28398 36 13 and and CCONJ bracis-28398 36 14 the the DET bracis-28398 36 15 meta meta NOUN bracis-28398 36 16 - - PUNCT bracis-28398 36 17 features feature NOUN bracis-28398 36 18 . . PUNCT bracis-28398 37 1 inspecting inspect VERB bracis-28398 37 2 general general ADJ bracis-28398 37 3 trends trend NOUN bracis-28398 37 4 in in ADP bracis-28398 37 5 the the DET bracis-28398 37 6 input input NOUN bracis-28398 37 7 features feature NOUN bracis-28398 37 8 of of ADP bracis-28398 37 9 the the DET bracis-28398 37 10 hard hard ADJ bracis-28398 37 11 instances instance NOUN bracis-28398 37 12 of of ADP bracis-28398 37 13 a a DET bracis-28398 37 14 dataset dataset NOUN bracis-28398 37 15 can can AUX bracis-28398 37 16 be be AUX bracis-28398 37 17 particularly particularly ADV bracis-28398 37 18 informative informative ADJ bracis-28398 37 19 and and CCONJ bracis-28398 37 20 allow allow VERB bracis-28398 37 21 to to PART bracis-28398 37 22 involve involve VERB bracis-28398 37 23 the the DET bracis-28398 37 24 data datum NOUN bracis-28398 37 25 domain domain NOUN bracis-28398 37 26 experts expert NOUN bracis-28398 37 27 in in ADP bracis-28398 37 28 the the DET bracis-28398 37 29 ml ml PROPN bracis-28398 37 30 pipeline pipeline NOUN bracis-28398 37 31 . . PUNCT bracis-28398 38 1 on on ADP bracis-28398 38 2 the the DET bracis-28398 38 3 other other ADJ bracis-28398 38 4 hand hand NOUN bracis-28398 38 5 , , PUNCT bracis-28398 38 6 looking look VERB bracis-28398 38 7 at at ADP bracis-28398 38 8 the the DET bracis-28398 38 9 hardness hardness NOUN bracis-28398 38 10 meta meta X bracis-28398 38 11 - - PUNCT bracis-28398 38 12 features feature NOUN bracis-28398 38 13 allows allow VERB bracis-28398 38 14 data datum NOUN bracis-28398 38 15 scientists scientist NOUN bracis-28398 38 16 to to PART bracis-28398 38 17 understand understand VERB bracis-28398 38 18 structural structural ADJ bracis-28398 38 19 problems problem NOUN bracis-28398 38 20 with with ADP bracis-28398 38 21 such such ADJ bracis-28398 38 22 instances instance NOUN bracis-28398 38 23 , , PUNCT bracis-28398 38 24 which which PRON bracis-28398 38 25 may may AUX bracis-28398 38 26 be be AUX bracis-28398 38 27 preventing prevent VERB bracis-28398 38 28 the the DET bracis-28398 38 29 obtainment obtainment NOUN bracis-28398 38 30 of of ADP bracis-28398 38 31 a a DET bracis-28398 38 32 better well ADJ bracis-28398 38 33 predictive predictive ADJ bracis-28398 38 34 accuracy accuracy NOUN bracis-28398 38 35 . . PUNCT bracis-28398 39 1 this this DET bracis-28398 39 2 dual dual ADJ bracis-28398 39 3 interplay interplay NOUN bracis-28398 39 4 allows allow VERB bracis-28398 39 5 for for ADP bracis-28398 39 6 taking take VERB bracis-28398 39 7 into into ADP bracis-28398 39 8 account account NOUN bracis-28398 39 9 the the DET bracis-28398 39 10 main main ADJ bracis-28398 39 11 stakeholders stakeholder NOUN bracis-28398 39 12 involved involve VERB bracis-28398 39 13 in in ADP bracis-28398 39 14 data datum NOUN bracis-28398 39 15 analysis analysis NOUN bracis-28398 39 16 for for ADP bracis-28398 39 17 decision decision NOUN bracis-28398 39 18 support support NOUN bracis-28398 39 19 . . PUNCT bracis-28398 40 1 in in ADP bracis-28398 40 2 particular particular ADJ bracis-28398 40 3 , , PUNCT bracis-28398 40 4 our our PRON bracis-28398 40 5 framework framework NOUN bracis-28398 40 6 is be AUX bracis-28398 40 7 applied apply VERB bracis-28398 40 8 here here ADV bracis-28398 40 9 to to ADP bracis-28398 40 10 datasets dataset NOUN bracis-28398 40 11 from from ADP bracis-28398 40 12 the the DET bracis-28398 40 13 health health NOUN bracis-28398 40 14 domain domain NOUN bracis-28398 40 15 , , PUNCT bracis-28398 40 16 which which PRON bracis-28398 40 17 is be AUX bracis-28398 40 18 a a DET bracis-28398 40 19 critically critically ADV bracis-28398 40 20 requester requester ADJ bracis-28398 40 21 reliable reliable ADJ bracis-28398 40 22 predictive predictive ADJ bracis-28398 40 23 models model NOUN bracis-28398 40 24 . . PUNCT bracis-28398 41 1 as as ADP bracis-28398 41 2 case case NOUN bracis-28398 41 3 studies study NOUN bracis-28398 41 4 , , PUNCT bracis-28398 41 5 we we PRON bracis-28398 41 6 consider consider VERB bracis-28398 41 7 a a DET bracis-28398 41 8 set set NOUN bracis-28398 41 9 of of ADP bracis-28398 41 10 covid covid NOUN bracis-28398 41 11 prognosis prognosis NOUN bracis-28398 41 12 datasets dataset NOUN bracis-28398 41 13 from from ADP bracis-28398 41 14 brazil brazil PROPN bracis-28398 41 15 , , PUNCT bracis-28398 41 16 one one NUM bracis-28398 41 17 of of ADP bracis-28398 41 18 the the DET bracis-28398 41 19 countries country NOUN bracis-28398 41 20 hit hit VERB bracis-28398 41 21 hardest hard ADV bracis-28398 41 22 by by ADP bracis-28398 41 23 the the DET bracis-28398 41 24 pandemic pandemic NOUN bracis-28398 41 25 . . PUNCT bracis-28398 42 1 the the DET bracis-28398 42 2 classification classification NOUN bracis-28398 42 3 problem problem NOUN bracis-28398 42 4 consists consist VERB bracis-28398 42 5 in in ADP bracis-28398 42 6 predicting predict VERB bracis-28398 42 7 whether whether SCONJ bracis-28398 42 8 a a DET bracis-28398 42 9 covid covid NOUN bracis-28398 42 10 - - PUNCT bracis-28398 42 11 hospitalized hospitalize VERB bracis-28398 42 12 patient patient NOUN bracis-28398 42 13 will will AUX bracis-28398 42 14 develop develop VERB bracis-28398 42 15 into into ADP bracis-28398 42 16 an an DET bracis-28398 42 17 aggravated aggravate VERB bracis-28398 42 18 condition condition NOUN bracis-28398 42 19 or or CCONJ bracis-28398 42 20 not not PART bracis-28398 42 21 , , PUNCT bracis-28398 42 22 information information NOUN bracis-28398 42 23 that that PRON bracis-28398 42 24 can can AUX bracis-28398 42 25 support support VERB bracis-28398 42 26 decision decision NOUN bracis-28398 42 27 making making NOUN bracis-28398 42 28 at at ADP bracis-28398 42 29 both both CCONJ bracis-28398 42 30 clinical clinical ADJ bracis-28398 42 31 and and CCONJ bracis-28398 42 32 management management NOUN bracis-28398 42 33 levels level NOUN bracis-28398 42 34 at at ADP bracis-28398 42 35 hospitals hospital NOUN bracis-28398 42 36 . . PUNCT bracis-28398 43 1 our our PRON bracis-28398 43 2 framework framework NOUN bracis-28398 43 3 contributes contribute VERB bracis-28398 43 4 on on ADP bracis-28398 43 5 the the DET bracis-28398 43 6 following following ADJ bracis-28398 43 7 ways way NOUN bracis-28398 43 8 : : PUNCT bracis-28398 43 9 providing provide VERB bracis-28398 43 10 a a DET bracis-28398 43 11 principled principled ADJ bracis-28398 43 12 approach approach NOUN bracis-28398 43 13 for for ADP bracis-28398 43 14 identifying identify VERB bracis-28398 43 15 hard hard ADJ bracis-28398 43 16 instances instance NOUN bracis-28398 43 17 in in ADP bracis-28398 43 18 a a DET bracis-28398 43 19 dataset dataset NOUN bracis-28398 43 20 : : PUNCT bracis-28398 43 21 taking take VERB bracis-28398 43 22 advantage advantage NOUN bracis-28398 43 23 of of ADP bracis-28398 43 24 the the DET bracis-28398 43 25 hardness hardness NOUN bracis-28398 43 26 embedding embedding NOUN bracis-28398 43 27 of of ADP bracis-28398 43 28 a a DET bracis-28398 43 29 dataset dataset NOUN bracis-28398 43 30 , , PUNCT bracis-28398 43 31 where where SCONJ bracis-28398 43 32 easy easy ADJ bracis-28398 43 33 and and CCONJ bracis-28398 43 34 hard hard ADJ bracis-28398 43 35 instances instance NOUN bracis-28398 43 36 are be AUX bracis-28398 43 37 interposed interpose VERB bracis-28398 43 38 , , PUNCT bracis-28398 43 39 different different ADJ bracis-28398 43 40 types type NOUN bracis-28398 43 41 of of ADP bracis-28398 43 42 analyzes analyze NOUN bracis-28398 43 43 are be AUX bracis-28398 43 44 made make VERB bracis-28398 43 45 possible possible ADJ bracis-28398 43 46 . . PUNCT bracis-28398 44 1 one one NUM bracis-28398 44 2 of of ADP bracis-28398 44 3 them they PRON bracis-28398 44 4 is be AUX bracis-28398 44 5 to to PART bracis-28398 44 6 define define VERB bracis-28398 44 7 hardness hardness NOUN bracis-28398 44 8 and and CCONJ bracis-28398 44 9 easiness easiness ADJ bracis-28398 44 10 footprints footprint NOUN bracis-28398 44 11 , , PUNCT bracis-28398 44 12 which which PRON bracis-28398 44 13 are be AUX bracis-28398 44 14 dense dense ADJ bracis-28398 44 15 regions region NOUN bracis-28398 44 16 of of ADP bracis-28398 44 17 the the DET bracis-28398 44 18 representation representation NOUN bracis-28398 44 19 space space NOUN bracis-28398 44 20 which which PRON bracis-28398 44 21 encompass encompass VERB bracis-28398 44 22 more more ADV bracis-28398 44 23 hard hard ADJ bracis-28398 44 24 and and CCONJ bracis-28398 44 25 easy easy ADJ bracis-28398 44 26 instances instance NOUN bracis-28398 44 27 , , PUNCT bracis-28398 44 28 respectively respectively ADV bracis-28398 44 29 . . PUNCT bracis-28398 45 1 this this PRON bracis-28398 45 2 allows allow VERB bracis-28398 45 3 to to PART bracis-28398 45 4 define define VERB bracis-28398 45 5 a a DET bracis-28398 45 6 principled principled ADJ bracis-28398 45 7 strategy strategy NOUN bracis-28398 45 8 for for ADP bracis-28398 45 9 choosing choose VERB bracis-28398 45 10 sets set NOUN bracis-28398 45 11 of of ADP bracis-28398 45 12 instances instance NOUN bracis-28398 45 13 to to PART bracis-28398 45 14 be be AUX bracis-28398 45 15 further far ADV bracis-28398 45 16 examined examine VERB bracis-28398 45 17 ; ; PUNCT bracis-28398 45 18 providing provide VERB bracis-28398 45 19 meaningful meaningful ADJ bracis-28398 45 20 insights insight NOUN bracis-28398 45 21 on on ADP bracis-28398 45 22 data datum NOUN bracis-28398 45 23 quality quality NOUN bracis-28398 45 24 and and CCONJ bracis-28398 45 25 structure structure NOUN bracis-28398 45 26 : : PUNCT bracis-28398 45 27 the the DET bracis-28398 45 28 hardness hardness ADJ bracis-28398 45 29 profile profile NOUN bracis-28398 45 30 of of ADP bracis-28398 45 31 a a DET bracis-28398 45 32 dataset dataset NOUN bracis-28398 45 33 can can AUX bracis-28398 45 34 be be AUX bracis-28398 45 35 inspected inspect VERB bracis-28398 45 36 for for ADP bracis-28398 45 37 gaining gain VERB bracis-28398 45 38 insights insight NOUN bracis-28398 45 39 about about ADP bracis-28398 45 40 where where SCONJ bracis-28398 45 41 and and CCONJ bracis-28398 45 42 why why SCONJ bracis-28398 45 43 ml ml NOUN bracis-28398 45 44 models model NOUN bracis-28398 45 45 are be AUX bracis-28398 45 46 failing fail VERB bracis-28398 45 47 ; ; PUNCT bracis-28398 45 48 being be AUX bracis-28398 45 49 actionable actionable ADJ bracis-28398 45 50 by by ADP bracis-28398 45 51 different different ADJ bracis-28398 45 52 stakeholders stakeholder NOUN bracis-28398 45 53 : : PUNCT bracis-28398 45 54 we we PRON bracis-28398 45 55 combine combine VERB bracis-28398 45 56 two two NUM bracis-28398 45 57 levels level NOUN bracis-28398 45 58 of of ADP bracis-28398 45 59 knowledge knowledge NOUN bracis-28398 45 60 . . PUNCT bracis-28398 46 1 while while SCONJ bracis-28398 46 2 inspecting inspect VERB bracis-28398 46 3 relationships relationship NOUN bracis-28398 46 4 of of ADP bracis-28398 46 5 the the DET bracis-28398 46 6 input input NOUN bracis-28398 46 7 features feature VERB bracis-28398 46 8 that that PRON bracis-28398 46 9 lead lead VERB bracis-28398 46 10 to to ADP bracis-28398 46 11 increased increase VERB bracis-28398 46 12 hardness hardness NOUN bracis-28398 46 13 levels level NOUN bracis-28398 46 14 can can AUX bracis-28398 46 15 be be AUX bracis-28398 46 16 a a DET bracis-28398 46 17 valuable valuable ADJ bracis-28398 46 18 tool tool NOUN bracis-28398 46 19 for for ADP bracis-28398 46 20 better well ADJ bracis-28398 46 21 data datum NOUN bracis-28398 46 22 understanding understanding NOUN bracis-28398 46 23 and and CCONJ bracis-28398 46 24 auditing auditing NOUN bracis-28398 46 25 with with ADP bracis-28398 46 26 the the DET bracis-28398 46 27 aid aid NOUN bracis-28398 46 28 of of ADP bracis-28398 46 29 domain domain NOUN bracis-28398 46 30 experts expert NOUN bracis-28398 46 31 , , PUNCT bracis-28398 46 32 the the DET bracis-28398 46 33 meta meta NOUN bracis-28398 46 34 - - PUNCT bracis-28398 46 35 features feature NOUN bracis-28398 46 36 allow allow VERB bracis-28398 46 37 to to PART bracis-28398 46 38 prospect prospect VERB bracis-28398 46 39 structural structural ADJ bracis-28398 46 40 problems problem NOUN bracis-28398 46 41 in in ADP bracis-28398 46 42 the the DET bracis-28398 46 43 data datum NOUN bracis-28398 46 44 which which PRON bracis-28398 46 45 can can AUX bracis-28398 46 46 be be AUX bracis-28398 46 47 informative informative ADJ bracis-28398 46 48 for for ADP bracis-28398 46 49 data data NOUN bracis-28398 46 50 science science NOUN bracis-28398 46 51 and and CCONJ bracis-28398 46 52 ml ml NOUN bracis-28398 46 53 practitioners practitioner NOUN bracis-28398 46 54 . . PUNCT bracis-28398 47 1 this this DET bracis-28398 47 2 paper paper NOUN bracis-28398 47 3 is be AUX bracis-28398 47 4 structured structure VERB bracis-28398 47 5 as as SCONJ bracis-28398 47 6 follows follow VERB bracis-28398 47 7 : : PUNCT bracis-28398 47 8 section section NOUN bracis-28398 47 9     SPACE bracis-28398 47 10 2 2 NUM bracis-28398 47 11 presents present VERB bracis-28398 47 12 some some DET bracis-28398 47 13 related related ADJ bracis-28398 47 14 work work NOUN bracis-28398 47 15 . . PUNCT bracis-28398 48 1 section section NOUN bracis-28398 48 2     SPACE bracis-28398 48 3 3 3 NUM bracis-28398 48 4 presents present VERB bracis-28398 48 5 our our PRON bracis-28398 48 6 formulation formulation NOUN bracis-28398 48 7 . . PUNCT bracis-28398 49 1 section section NOUN bracis-28398 49 2     SPACE bracis-28398 49 3 4 4 NUM bracis-28398 49 4 presents present VERB bracis-28398 49 5 some some DET bracis-28398 49 6 experimental experimental ADJ bracis-28398 49 7 results result NOUN bracis-28398 49 8 . . PUNCT bracis-28398 50 1 section section NOUN bracis-28398 50 2     SPACE bracis-28398 50 3 5 5 NUM bracis-28398 50 4 concludes conclude VERB bracis-28398 50 5 the the DET bracis-28398 50 6 paper paper NOUN bracis-28398 50 7 with with ADP bracis-28398 50 8 some some DET bracis-28398 50 9 discussions discussion NOUN bracis-28398 50 10 . . PUNCT bracis-28398 51 1 2 2 NUM bracis-28398 51 2 related relate VERB bracis-28398 51 3 work work NOUN bracis-28398 51 4 ml ml NOUN bracis-28398 51 5 techniques technique NOUN bracis-28398 51 6 are be AUX bracis-28398 51 7 data datum NOUN bracis-28398 51 8 - - PUNCT bracis-28398 51 9 hungry hungry ADJ bracis-28398 51 10 and and CCONJ bracis-28398 51 11 a a DET bracis-28398 51 12 common common ADJ bracis-28398 51 13 mistake mistake NOUN bracis-28398 51 14 is be AUX bracis-28398 51 15 to to PART bracis-28398 51 16 consider consider VERB bracis-28398 51 17 that that SCONJ bracis-28398 51 18 more more ADJ bracis-28398 51 19 data datum NOUN bracis-28398 51 20 will will AUX bracis-28398 51 21 always always ADV bracis-28398 51 22 result result VERB bracis-28398 51 23 in in ADP bracis-28398 51 24 more more ADV bracis-28398 51 25 accurate accurate ADJ bracis-28398 51 26 predictive predictive ADJ bracis-28398 51 27 models model NOUN bracis-28398 51 28 . . PUNCT bracis-28398 52 1 but but CCONJ bracis-28398 52 2 some some DET bracis-28398 52 3 issues issue NOUN bracis-28398 52 4 present present ADJ bracis-28398 52 5 in in ADP bracis-28398 52 6 real real ADJ bracis-28398 52 7 data datum NOUN bracis-28398 52 8 , , PUNCT bracis-28398 52 9 such such ADJ bracis-28398 52 10 as as ADP bracis-28398 52 11 noise noise NOUN bracis-28398 52 12 , , PUNCT bracis-28398 52 13 sparse sparse ADJ bracis-28398 52 14 regions region NOUN bracis-28398 52 15 and and CCONJ bracis-28398 52 16 outliers outlier NOUN bracis-28398 52 17 , , PUNCT bracis-28398 52 18 can can AUX bracis-28398 52 19 impair impair VERB bracis-28398 52 20 the the DET bracis-28398 52 21 ml ml NOUN bracis-28398 52 22 system system NOUN bracis-28398 52 23 ’s ’s PART bracis-28398 52 24 performance performance NOUN bracis-28398 52 25 , , PUNCT bracis-28398 52 26 trustfulness trustfulness NOUN bracis-28398 52 27 and and CCONJ bracis-28398 52 28 acceptance acceptance NOUN bracis-28398 52 29 . . PUNCT bracis-28398 53 1 therefore therefore ADV bracis-28398 53 2 , , PUNCT bracis-28398 53 3 data datum NOUN bracis-28398 53 4 quality quality NOUN bracis-28398 53 5 plays play VERB bracis-28398 53 6 a a DET bracis-28398 53 7 central central ADJ bracis-28398 53 8 role role NOUN bracis-28398 53 9 in in ADP bracis-28398 53 10 developing develop VERB bracis-28398 53 11 reliable reliable ADJ bracis-28398 53 12 ml ml NOUN bracis-28398 53 13 systems system NOUN bracis-28398 53 14 . . PUNCT bracis-28398 54 1 since since SCONJ bracis-28398 54 2 predictions prediction NOUN bracis-28398 54 3 inside inside ADP bracis-28398 54 4 health health NOUN bracis-28398 54 5 scenarios scenario NOUN bracis-28398 54 6 involve involve VERB bracis-28398 54 7 highly highly ADV bracis-28398 54 8 delicate delicate ADJ bracis-28398 54 9 matters matter NOUN bracis-28398 54 10 , , PUNCT bracis-28398 54 11 data datum NOUN bracis-28398 54 12 quality quality NOUN bracis-28398 54 13 assessment assessment NOUN bracis-28398 54 14 and and CCONJ bracis-28398 54 15 cleaning cleaning NOUN bracis-28398 54 16 are be AUX bracis-28398 54 17 even even ADV bracis-28398 54 18 more more ADV bracis-28398 54 19 critical critical ADJ bracis-28398 54 20 . . PUNCT bracis-28398 55 1 some some DET bracis-28398 55 2 recent recent ADJ bracis-28398 55 3 works work NOUN bracis-28398 55 4 have have AUX bracis-28398 55 5 used use VERB bracis-28398 55 6 data data NOUN bracis-28398 55 7 - - PUNCT bracis-28398 55 8 centered center VERB bracis-28398 55 9 approaches approach NOUN bracis-28398 55 10 to to PART bracis-28398 55 11 formulate formulate VERB bracis-28398 55 12 strategies strategy NOUN bracis-28398 55 13 allowing allow VERB bracis-28398 55 14 to to PART bracis-28398 55 15 : : PUNCT bracis-28398 55 16 assess assess VERB bracis-28398 55 17 how how SCONJ bracis-28398 55 18 difficult difficult ADJ bracis-28398 55 19 it it PRON bracis-28398 55 20 is be AUX bracis-28398 55 21 to to PART bracis-28398 55 22 predict predict VERB bracis-28398 55 23 the the DET bracis-28398 55 24 class class NOUN bracis-28398 55 25 of of ADP bracis-28398 55 26 each each DET bracis-28398 55 27 instance instance NOUN bracis-28398 55 28 in in ADP bracis-28398 55 29 a a DET bracis-28398 55 30 dataset dataset NOUN bracis-28398 55 31     SPACE bracis-28398 55 32 [ [ X bracis-28398 55 33 6 6 NUM bracis-28398 55 34 , , PUNCT bracis-28398 55 35 8 8 NUM bracis-28398 55 36 , , PUNCT bracis-28398 55 37 13 13 NUM bracis-28398 55 38 ] ] PUNCT bracis-28398 55 39 ; ; PUNCT bracis-28398 55 40 understand understand VERB bracis-28398 55 41 why why SCONJ bracis-28398 55 42 some some DET bracis-28398 55 43 instances instance NOUN bracis-28398 55 44 are be AUX bracis-28398 55 45 frequently frequently ADV bracis-28398 55 46 misclassified misclassifie VERB bracis-28398 55 47 , , PUNCT bracis-28398 55 48 while while SCONJ bracis-28398 55 49 also also ADV bracis-28398 55 50 trying try VERB bracis-28398 55 51 to to PART bracis-28398 55 52 explain explain VERB bracis-28398 55 53 the the DET bracis-28398 55 54 behavior behavior NOUN bracis-28398 55 55 of of ADP bracis-28398 55 56 classifiers classifier NOUN bracis-28398 55 57 for for ADP bracis-28398 55 58 such such ADJ bracis-28398 55 59 instances instance NOUN bracis-28398 55 60     SPACE bracis-28398 55 61 [ [ X bracis-28398 55 62 8 8 NUM bracis-28398 55 63 , , PUNCT bracis-28398 55 64 14 14 NUM bracis-28398 55 65 ] ] PUNCT bracis-28398 55 66 ; ; PUNCT bracis-28398 55 67 predict predict VERB bracis-28398 55 68 test test NOUN bracis-28398 55 69 instances instance NOUN bracis-28398 55 70 that that PRON bracis-28398 55 71 will will AUX bracis-28398 55 72 be be AUX bracis-28398 55 73 reliably reliably ADV bracis-28398 55 74 classified classified ADJ bracis-28398 55 75 or or CCONJ bracis-28398 56 1 not not PART bracis-28398 56 2     SPACE bracis-28398 56 3 [ [ X bracis-28398 56 4 7 7 NUM bracis-28398 56 5 ] ] PUNCT bracis-28398 56 6 ; ; PUNCT bracis-28398 56 7 remove remove VERB bracis-28398 56 8 hard hard ADJ bracis-28398 56 9 or or CCONJ bracis-28398 56 10 ambiguous ambiguous ADJ bracis-28398 56 11 instances instance NOUN bracis-28398 56 12 in in ADP bracis-28398 56 13 a a DET bracis-28398 56 14 data data NOUN bracis-28398 56 15 sculpting sculpt VERB bracis-28398 56 16 strategy strategy NOUN bracis-28398 56 17 to to PART bracis-28398 56 18 improve improve VERB bracis-28398 56 19 ml ml ADP bracis-28398 56 20 algorithms algorithms NOUN bracis-28398 56 21 performance performance NOUN bracis-28398 56 22     SPACE bracis-28398 57 1 [ [ X bracis-28398 57 2 6 6 NUM bracis-28398 57 3 , , PUNCT bracis-28398 57 4 13 13 NUM bracis-28398 57 5 ] ] PUNCT bracis-28398 57 6 ; ; PUNCT bracis-28398 57 7 and and CCONJ bracis-28398 57 8 identifying identify VERB bracis-28398 57 9 anomalous anomalous ADJ bracis-28398 57 10 regions region NOUN bracis-28398 57 11 of of ADP bracis-28398 57 12 the the DET bracis-28398 57 13 input input NOUN bracis-28398 57 14 feature feature NOUN bracis-28398 57 15 space space NOUN bracis-28398 57 16 offering offer VERB bracis-28398 57 17 useful useful ADJ bracis-28398 57 18 explanations explanation NOUN bracis-28398 57 19 and and CCONJ bracis-28398 57 20 insights insight NOUN bracis-28398 57 21 about about ADP bracis-28398 57 22 data datum NOUN bracis-28398 57 23 quality quality NOUN bracis-28398 57 24 as as ADV bracis-28398 57 25 well well ADV bracis-28398 57 26 as as ADP bracis-28398 57 27 model model NOUN bracis-28398 57 28 performance performance NOUN bracis-28398 57 29     SPACE bracis-28398 58 1 [ [ X bracis-28398 58 2 6 6 NUM bracis-28398 58 3 , , PUNCT bracis-28398 58 4 8 8 NUM bracis-28398 58 5 , , PUNCT bracis-28398 58 6 15 15 NUM bracis-28398 58 7 ] ] PUNCT bracis-28398 58 8 . . PUNCT bracis-28398 59 1 but but CCONJ bracis-28398 59 2 there there PRON bracis-28398 59 3 are be VERB bracis-28398 59 4 still still ADV bracis-28398 59 5 gaps gap NOUN bracis-28398 59 6 to to PART bracis-28398 59 7 be be AUX bracis-28398 59 8 filled fill VERB bracis-28398 59 9 that that PRON bracis-28398 59 10 may may AUX bracis-28398 59 11 help help VERB bracis-28398 59 12 increasing increase VERB bracis-28398 59 13 the the DET bracis-28398 59 14 acceptance acceptance NOUN bracis-28398 59 15 of of ADP bracis-28398 59 16 such such ADJ bracis-28398 59 17 tools tool NOUN bracis-28398 59 18 . . PUNCT bracis-28398 60 1 one one NUM bracis-28398 60 2 of of ADP bracis-28398 60 3 them they PRON bracis-28398 60 4 is be AUX bracis-28398 60 5 providing provide VERB bracis-28398 60 6 more more ADV bracis-28398 60 7 interpretable interpretable ADJ bracis-28398 60 8 and and CCONJ bracis-28398 60 9 actionable actionable ADJ bracis-28398 60 10 insights insight NOUN bracis-28398 60 11 to to ADP bracis-28398 60 12 the the DET bracis-28398 60 13 different different ADJ bracis-28398 60 14 stakeholders stakeholder NOUN bracis-28398 60 15 involved involve VERB bracis-28398 60 16 in in ADP bracis-28398 60 17 data datum NOUN bracis-28398 60 18 analysis analysis NOUN bracis-28398 60 19 . . PUNCT bracis-28398 61 1 in in ADP bracis-28398 61 2 order order NOUN bracis-28398 61 3 to to PART bracis-28398 61 4 increase increase VERB bracis-28398 61 5 the the DET bracis-28398 61 6 trustfulness trustfulness NOUN bracis-28398 61 7 and and CCONJ bracis-28398 61 8 acceptance acceptance NOUN bracis-28398 61 9 of of ADP bracis-28398 61 10 ai ai PROPN bracis-28398 61 11 systems system NOUN bracis-28398 61 12 , , PUNCT bracis-28398 61 13 it it PRON bracis-28398 61 14 is be AUX bracis-28398 61 15 imperative imperative ADJ bracis-28398 61 16 that that SCONJ bracis-28398 61 17 their their PRON bracis-28398 61 18 decisions decision NOUN bracis-28398 61 19 are be AUX bracis-28398 61 20 transparent transparent ADJ bracis-28398 61 21 and and CCONJ bracis-28398 61 22 can can AUX bracis-28398 61 23 be be AUX bracis-28398 61 24 scrutinized scrutinize VERB bracis-28398 61 25 by by ADP bracis-28398 61 26 different different ADJ bracis-28398 61 27 stakeholders stakeholder NOUN bracis-28398 61 28 . . PUNCT bracis-28398 62 1 this this PRON bracis-28398 62 2 has have AUX bracis-28398 62 3 lead lead VERB bracis-28398 62 4 to to ADP bracis-28398 62 5 the the DET bracis-28398 62 6 increase increase NOUN bracis-28398 62 7 in in ADP bracis-28398 62 8 the the DET bracis-28398 62 9 explainable explainable ADJ bracis-28398 62 10 artificial artificial ADJ bracis-28398 62 11 intelligence intelligence NOUN bracis-28398 62 12 ( ( PUNCT bracis-28398 62 13 xai xai PROPN bracis-28398 62 14 ) ) PUNCT bracis-28398 62 15 area area NOUN bracis-28398 62 16     SPACE bracis-28398 63 1 [ [ X bracis-28398 63 2 16 16 NUM bracis-28398 63 3 ] ] PUNCT bracis-28398 63 4 . . PUNCT bracis-28398 64 1 many many ADJ bracis-28398 64 2 xai xai PROPN bracis-28398 64 3 platforms platform NOUN bracis-28398 64 4 and and CCONJ bracis-28398 64 5 strategies strategy NOUN bracis-28398 64 6 are be AUX bracis-28398 64 7 currently currently ADV bracis-28398 64 8 available available ADJ bracis-28398 64 9 and and CCONJ bracis-28398 64 10 most most ADJ bracis-28398 64 11 of of ADP bracis-28398 64 12 them they PRON bracis-28398 64 13 focus focus VERB bracis-28398 64 14 on on ADP bracis-28398 64 15 evaluating evaluate VERB bracis-28398 64 16 how how SCONJ bracis-28398 64 17 the the DET bracis-28398 64 18 predictive predictive ADJ bracis-28398 64 19 input input NOUN bracis-28398 64 20 features feature VERB bracis-28398 64 21 influence influence VERB bracis-28398 64 22 the the DET bracis-28398 64 23 results result NOUN bracis-28398 64 24 of of ADP bracis-28398 64 25 the the DET bracis-28398 64 26 ml ml NOUN bracis-28398 64 27 models model NOUN bracis-28398 64 28 , , PUNCT bracis-28398 64 29 such such ADJ bracis-28398 64 30 as as ADP bracis-28398 64 31 permutation permutation NOUN bracis-28398 64 32 tests test NOUN bracis-28398 64 33     SPACE bracis-28398 65 1 [ [ X bracis-28398 65 2 17 17 NUM bracis-28398 65 3 ] ] PUNCT bracis-28398 65 4 and and CCONJ bracis-28398 65 5 shapley shapley NOUN bracis-28398 65 6 values value NOUN bracis-28398 65 7     SPACE bracis-28398 66 1 [ [ X bracis-28398 66 2 18 18 NUM bracis-28398 66 3 ] ] PUNCT bracis-28398 66 4 . . PUNCT bracis-28398 67 1 but but CCONJ bracis-28398 67 2 there there PRON bracis-28398 67 3 are be VERB bracis-28398 67 4 other other ADJ bracis-28398 67 5 layers layer NOUN bracis-28398 67 6 of of ADP bracis-28398 67 7 explainability explainability NOUN bracis-28398 67 8 that that PRON bracis-28398 67 9 are be AUX bracis-28398 67 10 not not PART bracis-28398 67 11 so so ADV bracis-28398 67 12 commonly commonly ADV bracis-28398 67 13 explored explore VERB bracis-28398 67 14 in in ADP bracis-28398 67 15 the the DET bracis-28398 67 16 literature literature NOUN bracis-28398 67 17 , , PUNCT bracis-28398 67 18 despite despite SCONJ bracis-28398 67 19 having have VERB bracis-28398 67 20 potential potential NOUN bracis-28398 67 21 to to PART bracis-28398 67 22 offer offer VERB bracis-28398 67 23 rich rich ADJ bracis-28398 67 24 insights insight NOUN bracis-28398 67 25 for for ADP bracis-28398 67 26 better well ADJ bracis-28398 67 27 understanding understand VERB bracis-28398 67 28 the the DET bracis-28398 67 29 strengths strength NOUN bracis-28398 67 30 and and CCONJ bracis-28398 67 31 weaknesses weakness NOUN bracis-28398 67 32 of of ADP bracis-28398 67 33 the the DET bracis-28398 67 34 ml ml PROPN bracis-28398 67 35 models model NOUN bracis-28398 67 36 . . PUNCT bracis-28398 68 1 one one NUM bracis-28398 68 2 of of ADP bracis-28398 68 3 them they PRON bracis-28398 68 4 is be AUX bracis-28398 68 5 working work VERB bracis-28398 68 6 at at ADP bracis-28398 68 7 a a DET bracis-28398 68 8 meta meta ADJ bracis-28398 68 9 - - PUNCT bracis-28398 68 10 level level NOUN bracis-28398 68 11 where where SCONJ bracis-28398 68 12 general general ADJ bracis-28398 68 13 properties property NOUN bracis-28398 68 14 of of ADP bracis-28398 68 15 the the DET bracis-28398 68 16 data datum NOUN bracis-28398 68 17 that that PRON bracis-28398 68 18 may may AUX bracis-28398 68 19 lead lead VERB bracis-28398 68 20 to to ADP bracis-28398 68 21 impairments impairment NOUN bracis-28398 68 22 in in ADP bracis-28398 68 23 a a DET bracis-28398 68 24 good good ADJ bracis-28398 68 25 predictive predictive ADJ bracis-28398 68 26 performance performance NOUN bracis-28398 68 27 are be AUX bracis-28398 68 28 identified identify VERB bracis-28398 68 29 and and CCONJ bracis-28398 68 30 characterized characterize VERB bracis-28398 68 31     SPACE bracis-28398 69 1 [ [ X bracis-28398 69 2 9 9 NUM bracis-28398 69 3 , , PUNCT bracis-28398 69 4 19 19 NUM bracis-28398 69 5 ] ] PUNCT bracis-28398 69 6 . . PUNCT bracis-28398 70 1 taking take VERB bracis-28398 70 2 steps step NOUN bracis-28398 70 3 towards towards ADP bracis-28398 70 4 this this DET bracis-28398 70 5 direction direction NOUN bracis-28398 70 6 , , PUNCT bracis-28398 70 7 we we PRON bracis-28398 70 8 can can AUX bracis-28398 70 9 extract extract VERB bracis-28398 70 10 local local ADJ bracis-28398 70 11 rules rule NOUN bracis-28398 70 12 based base VERB bracis-28398 70 13 on on ADP bracis-28398 70 14 at at ADP bracis-28398 70 15 most most ADV bracis-28398 70 16 two two NUM bracis-28398 70 17 input input NOUN bracis-28398 70 18 features feature NOUN bracis-28398 70 19 for for ADP bracis-28398 70 20 subsets subset NOUN bracis-28398 70 21 of of ADP bracis-28398 70 22 instances instance NOUN bracis-28398 70 23 a a DET bracis-28398 70 24 given give VERB bracis-28398 70 25 model model NOUN bracis-28398 70 26 struggles struggle NOUN bracis-28398 70 27 to to PART bracis-28398 70 28 classify classify VERB bracis-28398 70 29 correctly correctly ADV bracis-28398 70 30 , , PUNCT bracis-28398 70 31 that that ADV bracis-28398 70 32 is is ADV bracis-28398 70 33 , , PUNCT bracis-28398 70 34 instances instance VERB bracis-28398 70 35 the the DET bracis-28398 70 36 model model NOUN bracis-28398 70 37 regards regard VERB bracis-28398 70 38 as as ADV bracis-28398 70 39 hard hard ADJ bracis-28398 70 40 to to PART bracis-28398 70 41 classify classify VERB bracis-28398 70 42     SPACE bracis-28398 71 1 [ [ X bracis-28398 71 2 15 15 NUM bracis-28398 71 3 ] ] PUNCT bracis-28398 71 4 . . PUNCT bracis-28398 72 1 our our PRON bracis-28398 72 2 work work NOUN bracis-28398 72 3 follows follow VERB bracis-28398 72 4 a a DET bracis-28398 72 5 similar similar ADJ bracis-28398 72 6 direction direction NOUN bracis-28398 72 7 , , PUNCT bracis-28398 72 8 but but CCONJ bracis-28398 72 9 with with ADP bracis-28398 72 10 a a DET bracis-28398 72 11 more more ADV bracis-28398 72 12 general general ADJ bracis-28398 72 13 approach approach NOUN bracis-28398 72 14 for for ADP bracis-28398 72 15 finding find VERB bracis-28398 72 16 and and CCONJ bracis-28398 72 17 characterizing characterize VERB bracis-28398 72 18 the the DET bracis-28398 72 19 hard hard ADJ bracis-28398 72 20 instances instance NOUN bracis-28398 72 21 from from ADP bracis-28398 72 22 a a DET bracis-28398 72 23 dataset dataset NOUN bracis-28398 72 24 . . PUNCT bracis-28398 73 1 while while SCONJ bracis-28398 73 2 combining combine VERB bracis-28398 73 3 the the DET bracis-28398 73 4 outputs output NOUN bracis-28398 73 5 of of ADP bracis-28398 73 6 many many ADJ bracis-28398 73 7 models model NOUN bracis-28398 73 8 , , PUNCT bracis-28398 73 9 we we PRON bracis-28398 73 10 also also ADV bracis-28398 73 11 leverage leverage VERB bracis-28398 73 12 on on ADP bracis-28398 73 13 meta meta NOUN bracis-28398 73 14 - - PUNCT bracis-28398 73 15 features feature NOUN bracis-28398 73 16 able able ADJ bracis-28398 73 17 to to PART bracis-28398 73 18 describe describe VERB bracis-28398 73 19 why why SCONJ bracis-28398 73 20 some some DET bracis-28398 73 21 instances instance NOUN bracis-28398 73 22 are be AUX bracis-28398 73 23 hard hard ADJ bracis-28398 73 24 to to PART bracis-28398 73 25 classify classify VERB bracis-28398 73 26 , , PUNCT bracis-28398 73 27 giving give VERB bracis-28398 73 28 insights insight NOUN bracis-28398 73 29 at at ADP bracis-28398 73 30 different different ADJ bracis-28398 73 31 yet yet CCONJ bracis-28398 73 32 complementary complementary ADJ bracis-28398 73 33 perspectives perspective NOUN bracis-28398 73 34 . . PUNCT bracis-28398 74 1 3 3 NUM bracis-28398 74 2 formulation formulation NOUN bracis-28398 74 3 figure figure NOUN bracis-28398 74 4     SPACE bracis-28398 74 5 1 1 NUM bracis-28398 74 6 presents present VERB bracis-28398 74 7 an an DET bracis-28398 74 8 overview overview NOUN bracis-28398 74 9 of of ADP bracis-28398 74 10 the the DET bracis-28398 74 11 methodology methodology NOUN bracis-28398 74 12 followed follow VERB bracis-28398 74 13 in in ADP bracis-28398 74 14 our our PRON bracis-28398 74 15 paper paper NOUN bracis-28398 74 16 . . PUNCT bracis-28398 75 1 given give VERB bracis-28398 75 2 a a DET bracis-28398 75 3 dataset dataset VERB bracis-28398 75 4 \(\mathcal \(\mathcal ADJ bracis-28398 75 5 d\ d\ NOUN bracis-28398 75 6 ) ) PUNCT bracis-28398 75 7 , , PUNCT bracis-28398 75 8 a a DET bracis-28398 75 9 set set ADJ bracis-28398 75 10 \(\mathcal \(\mathcal ADJ bracis-28398 75 11 f\ f\ NOUN bracis-28398 75 12 ) ) PUNCT bracis-28398 75 13 of of ADP bracis-28398 75 14 hardness hardness ADJ bracis-28398 75 15 meta meta X bracis-28398 75 16 - - PUNCT bracis-28398 75 17 features feature NOUN bracis-28398 75 18 describing describe VERB bracis-28398 75 19 the the DET bracis-28398 75 20 level level NOUN bracis-28398 75 21 of of ADP bracis-28398 75 22 difficulty difficulty NOUN bracis-28398 75 23 in in ADP bracis-28398 75 24 classifying classify VERB bracis-28398 75 25 each each DET bracis-28398 75 26 instance instance NOUN bracis-28398 75 27 in in ADP bracis-28398 75 28 \(\mathcal \(\mathcal ADJ bracis-28398 75 29 d\ d\ NOUN bracis-28398 75 30 ) ) PUNCT bracis-28398 75 31 according accord VERB bracis-28398 75 32 to to ADP bracis-28398 75 33 different different ADJ bracis-28398 75 34 perspectives perspective NOUN bracis-28398 75 35 is be AUX bracis-28398 75 36 extracted extract VERB bracis-28398 75 37 . . PUNCT bracis-28398 76 1 algorithmic algorithmic ADJ bracis-28398 76 2 performance performance NOUN bracis-28398 76 3 \(\mathcal \(\mathcal ADJ bracis-28398 76 4 p\ p\ NOUN bracis-28398 76 5 ) ) PUNCT bracis-28398 76 6 is be AUX bracis-28398 76 7 registered register VERB bracis-28398 76 8 for for ADP bracis-28398 76 9 each each DET bracis-28398 76 10 individual individual ADJ bracis-28398 76 11 instance instance NOUN bracis-28398 76 12 of of ADP bracis-28398 76 13 the the DET bracis-28398 76 14 dataset dataset NOUN bracis-28398 76 15 for for ADP bracis-28398 76 16 a a DET bracis-28398 76 17 pool pool NOUN bracis-28398 76 18 of of ADP bracis-28398 76 19 classification classification NOUN bracis-28398 76 20 algorithms algorithm NOUN bracis-28398 76 21 \(\mathcal \(\mathcal NOUN bracis-28398 76 22 a\ a\ NOUN bracis-28398 76 23 ) ) PUNCT bracis-28398 76 24 . . PUNCT bracis-28398 77 1 joining join VERB bracis-28398 77 2 \(\mathcal \(\mathcal ADJ bracis-28398 77 3 f\ f\ NOUN bracis-28398 77 4 ) ) PUNCT bracis-28398 77 5 and and CCONJ bracis-28398 77 6 \(\mathcal \(\mathcal ADJ bracis-28398 77 7 p\ p\ NOUN bracis-28398 77 8 ) ) PUNCT bracis-28398 77 9 , , PUNCT bracis-28398 77 10 a a DET bracis-28398 77 11 hardness hardness NOUN bracis-28398 77 12 embedding embed VERB bracis-28398 77 13 is be AUX bracis-28398 77 14 produced produce VERB bracis-28398 77 15 where where SCONJ bracis-28398 77 16 the the DET bracis-28398 77 17 instances instance NOUN bracis-28398 77 18 are be AUX bracis-28398 77 19 placed place VERB bracis-28398 77 20 so so SCONJ bracis-28398 77 21 as as SCONJ bracis-28398 77 22 to to PART bracis-28398 77 23 present present VERB bracis-28398 77 24 linear linear ADJ bracis-28398 77 25 trends trend NOUN bracis-28398 77 26 of of ADP bracis-28398 77 27 difficulty difficulty NOUN bracis-28398 77 28 level level NOUN bracis-28398 77 29 , , PUNCT bracis-28398 77 30 from from ADP bracis-28398 77 31 bottom bottom ADJ bracis-28398 77 32 right right NOUN bracis-28398 77 33 ( ( PUNCT bracis-28398 77 34 easy easy ADJ bracis-28398 77 35 ) ) PUNCT bracis-28398 77 36 to to PART bracis-28398 77 37 top top VERB bracis-28398 77 38 left left ADJ bracis-28398 77 39 ( ( PUNCT bracis-28398 77 40 hard hard ADJ bracis-28398 77 41 ) ) PUNCT bracis-28398 77 42 . . PUNCT bracis-28398 78 1 next next ADJ bracis-28398 78 2 two two NUM bracis-28398 78 3 types type NOUN bracis-28398 78 4 of of ADP bracis-28398 78 5 analyzes analyze NOUN bracis-28398 78 6 are be AUX bracis-28398 78 7 performed perform VERB bracis-28398 78 8 . . PUNCT bracis-28398 79 1 one one NUM bracis-28398 79 2 of of ADP bracis-28398 79 3 them they PRON bracis-28398 79 4 involves involve VERB bracis-28398 79 5 defining define VERB bracis-28398 79 6 regions region NOUN bracis-28398 79 7 concentrating concentrate VERB bracis-28398 79 8 hard hard ADV bracis-28398 79 9 and and CCONJ bracis-28398 79 10 easy easy ADJ bracis-28398 79 11 to to PART bracis-28398 79 12 classify classify VERB bracis-28398 79 13 instances instance NOUN bracis-28398 79 14 . . PUNCT bracis-28398 80 1 they they PRON bracis-28398 80 2 are be AUX bracis-28398 80 3 called call VERB bracis-28398 80 4 hardness hardness ADJ bracis-28398 80 5 and and CCONJ bracis-28398 80 6 easiness easiness ADJ bracis-28398 80 7 footprints footprint NOUN bracis-28398 80 8 , , PUNCT bracis-28398 80 9 respectively respectively ADV bracis-28398 80 10 . . PUNCT bracis-28398 81 1 by by ADP bracis-28398 81 2 extracting extract VERB bracis-28398 81 3 patterns pattern NOUN bracis-28398 81 4 from from ADP bracis-28398 81 5 the the DET bracis-28398 81 6 opposing opposing ADJ bracis-28398 81 7 footprints footprint NOUN bracis-28398 81 8 , , PUNCT bracis-28398 81 9 we we PRON bracis-28398 81 10 are be AUX bracis-28398 81 11 able able ADJ bracis-28398 81 12 to to PART bracis-28398 81 13 obtain obtain VERB bracis-28398 81 14 insights insight NOUN bracis-28398 81 15 on on ADP bracis-28398 81 16 instance instance NOUN bracis-28398 81 17 hardness hardness NOUN bracis-28398 81 18 at at ADP bracis-28398 81 19 different different ADJ bracis-28398 81 20 levels level NOUN bracis-28398 81 21 and and CCONJ bracis-28398 81 22 perspectives perspective NOUN bracis-28398 81 23 . . PUNCT bracis-28398 82 1 the the DET bracis-28398 82 2 second second ADJ bracis-28398 82 3 approach approach NOUN bracis-28398 82 4 involves involve VERB bracis-28398 82 5 a a DET bracis-28398 82 6 visual visual ADJ bracis-28398 82 7 inspection inspection NOUN bracis-28398 82 8 of of ADP bracis-28398 82 9 the the DET bracis-28398 82 10 hardness hardness NOUN bracis-28398 82 11 embedding embed VERB bracis-28398 82 12 for for ADP bracis-28398 82 13 identifying identify VERB bracis-28398 82 14 observations observation NOUN bracis-28398 82 15 lying lie VERB bracis-28398 82 16 in in ADP bracis-28398 82 17 regions region NOUN bracis-28398 82 18 of of ADP bracis-28398 82 19 interest interest NOUN bracis-28398 82 20 to to PART bracis-28398 82 21 be be AUX bracis-28398 82 22 inspected inspect VERB bracis-28398 82 23 . . PUNCT bracis-28398 83 1 fig fig NOUN bracis-28398 83 2 . . PUNCT bracis-28398 84 1 1 1 X bracis-28398 84 2 . . X bracis-28398 84 3 overview overview NOUN bracis-28398 84 4 of of ADP bracis-28398 84 5 the the DET bracis-28398 84 6 proposed propose VERB bracis-28398 84 7 framework framework NOUN bracis-28398 84 8 . . PUNCT bracis-28398 85 1 full full ADJ bracis-28398 85 2 size size NOUN bracis-28398 85 3 image image NOUN bracis-28398 85 4 3.1 3.1 NUM bracis-28398 85 5 instance instance NOUN bracis-28398 85 6 hardness hardness NOUN bracis-28398 85 7 analysis analysis NOUN bracis-28398 85 8 let let VERB bracis-28398 85 9 us we PRON bracis-28398 85 10 define define VERB bracis-28398 85 11 \(\mathcal \(\mathcal ADJ bracis-28398 85 12 d\ d\ NOUN bracis-28398 85 13 ) ) PUNCT bracis-28398 85 14 formally formally ADV bracis-28398 85 15 as as ADP bracis-28398 85 16 containing contain VERB bracis-28398 85 17 n n DET bracis-28398 85 18 pairs pair NOUN bracis-28398 85 19 of of ADP bracis-28398 85 20 labeled label VERB bracis-28398 85 21 observations observation NOUN bracis-28398 85 22 ( ( PUNCT bracis-28398 85 23 \(\textbf{x}_i \(\textbf{x}_i PROPN bracis-28398 85 24 , , PUNCT bracis-28398 85 25 y_i y_i NUM bracis-28398 85 26 \ \ PROPN bracis-28398 85 27 ) ) PUNCT bracis-28398 85 28 ) ) PUNCT bracis-28398 85 29 . . PUNCT bracis-28398 86 1 each each DET bracis-28398 86 2 \(\textbf{x}_i \(\textbf{x}_i PROPN bracis-28398 86 3 \in \in PROPN bracis-28398 86 4 \mathcal \mathcal PROPN bracis-28398 86 5 x\ x\ NOUN bracis-28398 86 6 ) ) PUNCT bracis-28398 86 7 is be AUX bracis-28398 86 8 an an DET bracis-28398 86 9 instance instance NOUN bracis-28398 86 10 described describe VERB bracis-28398 86 11 by by ADP bracis-28398 86 12 m m NOUN bracis-28398 86 13 input input NOUN bracis-28398 86 14 features feature NOUN bracis-28398 86 15 and and CCONJ bracis-28398 86 16 is be AUX bracis-28398 86 17 labeled label VERB bracis-28398 86 18 in in ADP bracis-28398 86 19 \(y_i \(y_i NOUN bracis-28398 86 20 \in \in PROPN bracis-28398 86 21 \mathcal \mathcal PROPN bracis-28398 86 22 y\ y\ NOUN bracis-28398 86 23 ) ) PUNCT bracis-28398 86 24 , , PUNCT bracis-28398 86 25 where where SCONJ bracis-28398 86 26 \(\mathcal \(\mathcal ADJ bracis-28398 86 27 y\ y\ NOUN bracis-28398 86 28 ) ) PUNCT bracis-28398 86 29 is be AUX bracis-28398 86 30 a a DET bracis-28398 86 31 discrete discrete ADJ bracis-28398 86 32 and and CCONJ bracis-28398 86 33 non non ADJ bracis-28398 86 34 - - ADJ bracis-28398 86 35 ordered ordered ADJ bracis-28398 86 36 set set NOUN bracis-28398 86 37 of of ADP bracis-28398 86 38 classes class NOUN bracis-28398 86 39 . . PUNCT bracis-28398 87 1 in in ADP bracis-28398 87 2 addition addition NOUN bracis-28398 87 3 , , PUNCT bracis-28398 87 4 let let VERB bracis-28398 87 5 \(h \(h PROPN bracis-28398 87 6 : : PUNCT bracis-28398 87 7 \mathcal \mathcal PROPN bracis-28398 87 8 x x VERB bracis-28398 87 9 \rightarrow \rightarrow PROPN bracis-28398 87 10 \mathcal \mathcal PROPN bracis-28398 87 11 y\ y\ NOUN bracis-28398 87 12 ) ) PUNCT bracis-28398 87 13 denote denote VERB bracis-28398 87 14 a a DET bracis-28398 87 15 classification classification NOUN bracis-28398 87 16 hypothesis hypothesis NOUN bracis-28398 87 17 , , PUNCT bracis-28398 87 18 that that ADV bracis-28398 87 19 is is ADV bracis-28398 87 20 , , PUNCT bracis-28398 87 21 a a DET bracis-28398 87 22 ml ml ADP bracis-28398 87 23 predictive predictive ADJ bracis-28398 87 24 model model NOUN bracis-28398 87 25 generated generate VERB bracis-28398 87 26 from from ADP bracis-28398 87 27 \(\mathcal \(\mathcal ADJ bracis-28398 87 28 d\ d\ PROPN bracis-28398 87 29 ) ) PUNCT bracis-28398 87 30 . . PUNCT bracis-28398 88 1 the the DET bracis-28398 88 2 instance instance NOUN bracis-28398 88 3 hardness hardness NOUN bracis-28398 88 4 of of ADP bracis-28398 88 5 an an DET bracis-28398 88 6 instance instance NOUN bracis-28398 88 7 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 88 8 ) ) PUNCT bracis-28398 88 9 can can AUX bracis-28398 88 10 be be AUX bracis-28398 88 11 defined define VERB bracis-28398 88 12 as as ADP bracis-28398 88 13 the the DET bracis-28398 88 14 probability probability NOUN bracis-28398 88 15 of of ADP bracis-28398 88 16 misclassification misclassification NOUN bracis-28398 88 17 when when SCONJ bracis-28398 88 18 it it PRON bracis-28398 88 19 is be AUX bracis-28398 88 20 subject subject ADJ bracis-28398 88 21 to to ADP bracis-28398 88 22 a a DET bracis-28398 88 23 pool pool NOUN bracis-28398 88 24 of of ADP bracis-28398 88 25 different different ADJ bracis-28398 88 26 learning learning NOUN bracis-28398 88 27 algorithms algorithm NOUN bracis-28398 88 28 , , PUNCT bracis-28398 88 29 that that PRON bracis-28398 88 30 is be AUX bracis-28398 88 31 : : PUNCT bracis-28398 88 32 $ $ SYM bracis-28398 88 33 $ $ SYM bracis-28398 88 34 \begin{aligned \begin{aligne VERB bracis-28398 88 35 } } PUNCT bracis-28398 88 36 ih_{\mathcal ih_{\mathcal ADJ bracis-28398 88 37 { { PUNCT bracis-28398 88 38 a}}\big a}}\big PROPN bracis-28398 88 39 ( ( PUNCT bracis-28398 88 40 \textbf{x}_i \textbf{x}_i X bracis-28398 88 41 , , PUNCT bracis-28398 88 42 y_i y_i NUM bracis-28398 88 43 \big \big PROPN bracis-28398 88 44 ) ) PUNCT bracis-28398 89 1 = = SYM bracis-28398 89 2 1 1 NUM bracis-28398 89 3 \frac{1}{|\mathcal \frac{1}{|\mathcal ADJ bracis-28398 89 4 { { PUNCT bracis-28398 89 5 a}| a}| PROPN bracis-28398 89 6 } } PUNCT bracis-28398 89 7 \sum \sum NOUN bracis-28398 89 8 _ _ PUNCT bracis-28398 89 9 { { PUNCT bracis-28398 89 10 j=1}^{|\mathcal j=1}^{|\mathcal PROPN bracis-28398 89 11 { { PUNCT bracis-28398 89 12 a}| a}| PROPN bracis-28398 89 13 } } PUNCT bracis-28398 89 14 p\big p\big INTJ bracis-28398 89 15 ( ( PUNCT bracis-28398 89 16 y_i y_i NOUN bracis-28398 89 17 | | ADV bracis-28398 89 18 \textbf{x}_i \textbf{x}_i ADJ bracis-28398 89 19 , , PUNCT bracis-28398 89 20 h_j(\mathcal h_j(\mathcal ADJ bracis-28398 89 21 d)\big d)\big PROPN bracis-28398 89 22 ) ) PUNCT bracis-28398 89 23 , , PUNCT bracis-28398 89 24 \end{aligned}$$ \end{aligned}$$ X bracis-28398 89 25 ( ( PUNCT bracis-28398 89 26 1 1 NUM bracis-28398 89 27 ) ) PUNCT bracis-28398 89 28 where where SCONJ bracis-28398 89 29 \(p\big \(p\big PROPN bracis-28398 89 30 ( ( PUNCT bracis-28398 89 31 y_i y_i NOUN bracis-28398 89 32 | | ADV bracis-28398 89 33 \textbf{x}_i \textbf{x}_i ADJ bracis-28398 89 34 , , PUNCT bracis-28398 89 35 h_j(\mathcal h_j(\mathcal ADJ bracis-28398 89 36 d)\big d)\big PROPN bracis-28398 89 37 ) ) PUNCT bracis-28398 89 38 \ \ PROPN bracis-28398 89 39 ) ) PUNCT bracis-28398 89 40 is be AUX bracis-28398 89 41 the the DET bracis-28398 89 42 probability probability NOUN bracis-28398 89 43 the the DET bracis-28398 89 44 j j PROPN bracis-28398 89 45 - - PUNCT bracis-28398 89 46 th th X bracis-28398 89 47 ml ml NOUN bracis-28398 89 48 model model NOUN bracis-28398 89 49 in in ADP bracis-28398 89 50 the the DET bracis-28398 89 51 pool pool NOUN bracis-28398 89 52 attributes attribute VERB bracis-28398 89 53 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-28398 89 54 ) ) PUNCT bracis-28398 89 55 to to ADP bracis-28398 89 56 its its PRON bracis-28398 89 57 expected expect VERB bracis-28398 89 58 class class NOUN bracis-28398 89 59 \(y_i\ \(y_i\ NOUN bracis-28398 89 60 ) ) PUNCT bracis-28398 89 61 . . PUNCT bracis-28398 90 1 the the DET bracis-28398 90 2 intuition intuition NOUN bracis-28398 90 3 is be AUX bracis-28398 90 4 that that SCONJ bracis-28398 90 5 if if SCONJ bracis-28398 90 6 a a DET bracis-28398 90 7 pool pool NOUN bracis-28398 90 8 of of ADP bracis-28398 90 9 diverse diverse ADJ bracis-28398 90 10 classifiers classifier NOUN bracis-28398 90 11 is be AUX bracis-28398 90 12 unable unable ADJ bracis-28398 90 13 to to PART bracis-28398 90 14 predict predict VERB bracis-28398 90 15 the the DET bracis-28398 90 16 expected expect VERB bracis-28398 90 17 class class NOUN bracis-28398 90 18 of of ADP bracis-28398 90 19 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 90 20 ) ) PUNCT bracis-28398 90 21 with with ADP bracis-28398 90 22 a a DET bracis-28398 90 23 high high ADJ bracis-28398 90 24 probability probability NOUN bracis-28398 90 25 ( ( PUNCT bracis-28398 90 26 proxy proxy NOUN bracis-28398 90 27 to to ADP bracis-28398 90 28 a a DET bracis-28398 90 29 high high ADJ bracis-28398 90 30 confidence confidence NOUN bracis-28398 90 31 ) ) PUNCT bracis-28398 90 32 , , PUNCT bracis-28398 90 33 then then ADV bracis-28398 90 34 this this DET bracis-28398 90 35 instance instance NOUN bracis-28398 90 36 is be AUX bracis-28398 90 37 intrinsically intrinsically ADV bracis-28398 90 38 hard hard ADJ bracis-28398 90 39 to to PART bracis-28398 90 40 classify classify VERB bracis-28398 90 41 . . PUNCT bracis-28398 91 1 in in ADP bracis-28398 91 2 turn turn NOUN bracis-28398 91 3 , , PUNCT bracis-28398 91 4 hardness hardness ADJ bracis-28398 91 5 meta meta X bracis-28398 91 6 - - PUNCT bracis-28398 91 7 features feature NOUN bracis-28398 91 8 ( ( PUNCT bracis-28398 91 9 hm hm INTJ bracis-28398 91 10 ) ) PUNCT bracis-28398 91 11 can can AUX bracis-28398 91 12 be be AUX bracis-28398 91 13 used use VERB bracis-28398 91 14 for for ADP bracis-28398 91 15 prospecting prospect VERB bracis-28398 91 16 possible possible ADJ bracis-28398 91 17 reasons reason NOUN bracis-28398 91 18 why why SCONJ bracis-28398 91 19 an an DET bracis-28398 91 20 instance instance NOUN bracis-28398 91 21 is be AUX bracis-28398 91 22 hard hard ADJ bracis-28398 91 23 to to PART bracis-28398 91 24 classify classify VERB bracis-28398 91 25 or or CCONJ bracis-28398 92 1 not not PART bracis-28398 92 2     SPACE bracis-28398 92 3 [ [ X bracis-28398 92 4 9 9 NUM bracis-28398 92 5 , , PUNCT bracis-28398 92 6 10 10 NUM bracis-28398 92 7 ] ] PUNCT bracis-28398 92 8 . . PUNCT bracis-28398 93 1 summarizing summarize VERB bracis-28398 93 2 , , PUNCT bracis-28398 93 3 given give VERB bracis-28398 93 4 an an DET bracis-28398 93 5 instance instance NOUN bracis-28398 93 6 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 93 7 ): ): PUNCT bracis-28398 93 8 kdn kdn PROPN bracis-28398 93 9 ( ( PUNCT bracis-28398 93 10 k k ADJ bracis-28398 93 11 - - ADJ bracis-28398 93 12 disagreeing disagree VERB bracis-28398 93 13 neighbors neighbor NOUN bracis-28398 93 14 ) ) PUNCT bracis-28398 93 15 gives give VERB bracis-28398 93 16 the the DET bracis-28398 93 17 percentage percentage NOUN bracis-28398 93 18 of of ADP bracis-28398 93 19 k k PROPN bracis-28398 93 20 nearest near ADJ bracis-28398 93 21 neighbors neighbor NOUN bracis-28398 93 22 of of ADP bracis-28398 93 23 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 93 24 ) ) PUNCT bracis-28398 93 25 with with ADP bracis-28398 93 26 label label NOUN bracis-28398 93 27 different different ADJ bracis-28398 93 28 from from ADP bracis-28398 93 29 \(y_i\ \(y_i\ NOUN bracis-28398 93 30 ) ) PUNCT bracis-28398 93 31 ; ; PUNCT bracis-28398 93 32 n1i n1i PROPN bracis-28398 93 33 ( ( PUNCT bracis-28398 93 34 fraction fraction NOUN bracis-28398 93 35 of of ADP bracis-28398 93 36 nearby nearby ADJ bracis-28398 93 37 instances instance NOUN bracis-28398 93 38 of of ADP bracis-28398 93 39 different different ADJ bracis-28398 93 40 classes class NOUN bracis-28398 93 41 ) ) PUNCT bracis-28398 93 42 gives give VERB bracis-28398 93 43 the the DET bracis-28398 93 44 fraction fraction NOUN bracis-28398 93 45 of of ADP bracis-28398 93 46 instances instance NOUN bracis-28398 93 47 that that PRON bracis-28398 93 48 do do AUX bracis-28398 93 49 not not PART bracis-28398 93 50 belong belong VERB bracis-28398 93 51 to to ADP bracis-28398 93 52 class class NOUN bracis-28398 93 53 \(y_i\ \(y_i\ NOUN bracis-28398 93 54 ) ) PUNCT bracis-28398 93 55 which which PRON bracis-28398 93 56 are be AUX bracis-28398 93 57 connected connect VERB bracis-28398 93 58 to to ADP bracis-28398 93 59 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-28398 93 60 ) ) PUNCT bracis-28398 93 61 in in ADP bracis-28398 93 62 a a DET bracis-28398 93 63 minimum minimum ADJ bracis-28398 93 64 spanning span VERB bracis-28398 93 65 tree tree NOUN bracis-28398 93 66 built build VERB bracis-28398 93 67 from from ADP bracis-28398 93 68 \(\mathcal \(\mathcal ADJ bracis-28398 93 69 d\ d\ PROPN bracis-28398 93 70 ) ) PUNCT bracis-28398 93 71 ; ; PUNCT bracis-28398 93 72 n2i n2i PROPN bracis-28398 93 73 ( ( PUNCT bracis-28398 93 74 ratio ratio NOUN bracis-28398 93 75 of of ADP bracis-28398 93 76 the the DET bracis-28398 93 77 intra intra ADJ bracis-28398 93 78 - - ADJ bracis-28398 93 79 class class ADJ bracis-28398 93 80 and and CCONJ bracis-28398 93 81 extra extra ADJ bracis-28398 93 82 - - ADJ bracis-28398 93 83 class class ADJ bracis-28398 93 84 distances distance NOUN bracis-28398 93 85 ) ) PUNCT bracis-28398 93 86 takes take VERB bracis-28398 93 87 the the DET bracis-28398 93 88 ratio ratio NOUN bracis-28398 93 89 of of ADP bracis-28398 93 90 the the DET bracis-28398 93 91 distance distance NOUN bracis-28398 93 92 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 93 93 ) ) PUNCT bracis-28398 93 94 has have VERB bracis-28398 93 95 to to ADP bracis-28398 93 96 the the DET bracis-28398 93 97 nearest near ADJ bracis-28398 93 98 neighbor neighbor NOUN bracis-28398 93 99 from from ADP bracis-28398 93 100 its its PRON bracis-28398 93 101 class class NOUN bracis-28398 93 102 \(y_i\ \(y_i\ NOUN bracis-28398 93 103 ) ) PUNCT bracis-28398 93 104 to to ADP bracis-28398 93 105 the the DET bracis-28398 93 106 distance distance NOUN bracis-28398 93 107 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 93 108 ) ) PUNCT bracis-28398 93 109 has have VERB bracis-28398 93 110 to to ADP bracis-28398 93 111 its its PRON bracis-28398 93 112 nearest near ADJ bracis-28398 93 113 enemy enemy NOUN bracis-28398 93 114 ( ( PUNCT bracis-28398 93 115 nearest near ADJ bracis-28398 93 116 instance instance NOUN bracis-28398 93 117 from from ADP bracis-28398 93 118 another another DET bracis-28398 93 119 class class NOUN bracis-28398 93 120 \(y_j \(y_j VERB bracis-28398 93 121 \ne \ne PROPN bracis-28398 93 122 y_i\ y_i\ NOUN bracis-28398 93 123 ) ) PUNCT bracis-28398 93 124 ) ) PUNCT bracis-28398 93 125 ; ; PUNCT bracis-28398 93 126 lsci lsci PROPN bracis-28398 93 127 ( ( PUNCT bracis-28398 93 128 local local ADJ bracis-28398 93 129 set set VERB bracis-28398 93 130 cardinality cardinality NOUN bracis-28398 93 131 ) ) PUNCT bracis-28398 93 132 measures measure VERB bracis-28398 93 133 the the DET bracis-28398 93 134 size size NOUN bracis-28398 93 135 of of ADP bracis-28398 93 136 the the DET bracis-28398 93 137 set set NOUN bracis-28398 93 138 containing contain VERB bracis-28398 93 139 instances instance NOUN bracis-28398 93 140 from from ADP bracis-28398 93 141 class class NOUN bracis-28398 93 142 \(y_i\ \(y_i\ NOUN bracis-28398 93 143 ) ) PUNCT bracis-28398 93 144 which which PRON bracis-28398 93 145 are be AUX bracis-28398 93 146 closest close ADJ bracis-28398 93 147 to to ADP bracis-28398 93 148 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 93 149 ) ) PUNCT bracis-28398 93 150 than than ADP bracis-28398 93 151 to to ADP bracis-28398 93 152 its its PRON bracis-28398 93 153 nearest near ADJ bracis-28398 93 154 enemy enemy NOUN bracis-28398 93 155 ( ( PUNCT bracis-28398 93 156 this this DET bracis-28398 93 157 set set NOUN bracis-28398 93 158 is be AUX bracis-28398 93 159 named name VERB bracis-28398 93 160 local local ADJ bracis-28398 93 161 set set NOUN bracis-28398 93 162 of of ADP bracis-28398 93 163 the the DET bracis-28398 93 164 instance instance NOUN bracis-28398 93 165 ) ) PUNCT bracis-28398 93 166 ; ; PUNCT bracis-28398 93 167 lsr lsr PROPN bracis-28398 93 168 ( ( PUNCT bracis-28398 93 169 local local ADJ bracis-28398 93 170 set set VERB bracis-28398 93 171 radius radius NOUN bracis-28398 93 172 ) ) PUNCT bracis-28398 93 173 gives give VERB bracis-28398 93 174 the the DET bracis-28398 93 175 radius radius NOUN bracis-28398 93 176 of of ADP bracis-28398 93 177 the the DET bracis-28398 93 178 previous previous ADJ bracis-28398 93 179 set set NOUN bracis-28398 93 180 ; ; PUNCT bracis-28398 93 181 u u NOUN bracis-28398 93 182 ( ( PUNCT bracis-28398 93 183 usefulness usefulness NOUN bracis-28398 93 184 ) ) PUNCT bracis-28398 93 185 considers consider VERB bracis-28398 93 186 the the DET bracis-28398 93 187 number number NOUN bracis-28398 93 188 of of ADP bracis-28398 93 189 local local ADJ bracis-28398 93 190 sets set NOUN bracis-28398 93 191 an an DET bracis-28398 93 192 instance instance NOUN bracis-28398 93 193 belongs belong VERB bracis-28398 93 194 to to ADP bracis-28398 93 195 ; ; PUNCT bracis-28398 93 196 h h NOUN bracis-28398 93 197 ( ( PUNCT bracis-28398 93 198 harmfulness harmfulness NOUN bracis-28398 93 199 ) ) PUNCT bracis-28398 93 200 takes take VERB bracis-28398 93 201 the the DET bracis-28398 93 202 number number NOUN bracis-28398 93 203 of of ADP bracis-28398 93 204 instances instance NOUN bracis-28398 93 205 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-28398 93 206 ) ) PUNCT bracis-28398 93 207 is be AUX bracis-28398 93 208 nearest near ADJ bracis-28398 93 209 enemy enemy NOUN bracis-28398 93 210 of of ADP bracis-28398 93 211 ; ; PUNCT bracis-28398 93 212 cl cl INTJ bracis-28398 93 213 ( ( PUNCT bracis-28398 93 214 class class NOUN bracis-28398 93 215 likelihood likelihood NOUN bracis-28398 93 216 ) ) PUNCT bracis-28398 93 217 measures measure VERB bracis-28398 93 218 the the DET bracis-28398 93 219 likelihood likelihood NOUN bracis-28398 93 220 an an DET bracis-28398 93 221 instance instance NOUN bracis-28398 93 222 belongs belong VERB bracis-28398 93 223 to to ADP bracis-28398 93 224 its its PRON bracis-28398 93 225 class class NOUN bracis-28398 93 226 \(y_i\ \(y_i\ NOUN bracis-28398 93 227 ) ) PUNCT bracis-28398 93 228 ; ; PUNCT bracis-28398 93 229 cld cld NOUN bracis-28398 93 230 ( ( PUNCT bracis-28398 93 231 class class NOUN bracis-28398 93 232 likelihood likelihood NOUN bracis-28398 93 233 difference difference NOUN bracis-28398 93 234 ) ) PUNCT bracis-28398 93 235 measures measure VERB bracis-28398 93 236 the the DET bracis-28398 93 237 difference difference NOUN bracis-28398 93 238 between between ADP bracis-28398 93 239 the the DET bracis-28398 93 240 likelihood likelihood NOUN bracis-28398 93 241 an an DET bracis-28398 93 242 instance instance NOUN bracis-28398 93 243 belongs belong VERB bracis-28398 93 244 to to ADP bracis-28398 93 245 the the DET bracis-28398 93 246 class class NOUN bracis-28398 93 247 \(y_i\ \(y_i\ NOUN bracis-28398 93 248 ) ) PUNCT bracis-28398 93 249 to to ADP bracis-28398 93 250 the the DET bracis-28398 93 251 likelihood likelihood NOUN bracis-28398 93 252 it it PRON bracis-28398 93 253 belongs belong VERB bracis-28398 93 254 to to ADP bracis-28398 93 255 any any DET bracis-28398 93 256 other other ADJ bracis-28398 93 257 class class NOUN bracis-28398 93 258 \(y_j \(y_j NOUN bracis-28398 93 259 \ne \ne PROPN bracis-28398 93 260 y_i\ y_i\ NOUN bracis-28398 93 261 ) ) PUNCT bracis-28398 93 262 ; ; PUNCT bracis-28398 93 263 f1i f1i PROPN bracis-28398 93 264 ( ( PUNCT bracis-28398 93 265 fraction fraction NOUN bracis-28398 93 266 of of ADP bracis-28398 93 267 features feature NOUN bracis-28398 93 268 in in ADP bracis-28398 93 269 overlapping overlap VERB bracis-28398 93 270 areas area NOUN bracis-28398 93 271 ) ) PUNCT bracis-28398 93 272 gives give VERB bracis-28398 93 273 the the DET bracis-28398 93 274 percentage percentage NOUN bracis-28398 93 275 of of ADP bracis-28398 93 276 input input NOUN bracis-28398 93 277 features feature NOUN bracis-28398 93 278 lying lie VERB bracis-28398 93 279 in in ADP bracis-28398 93 280 overlapping overlap VERB bracis-28398 93 281 regions region NOUN bracis-28398 93 282 of of ADP bracis-28398 93 283 the the DET bracis-28398 93 284 classes class NOUN bracis-28398 93 285 ; ; PUNCT bracis-28398 93 286 dcp dcp NOUN bracis-28398 93 287 ( ( PUNCT bracis-28398 93 288 disjunct disjunct ADJ bracis-28398 93 289 class class NOUN bracis-28398 93 290 percentage percentage NOUN bracis-28398 93 291 ) ) PUNCT bracis-28398 93 292 gives give VERB bracis-28398 93 293 the the DET bracis-28398 93 294 percentage percentage NOUN bracis-28398 93 295 of of ADP bracis-28398 93 296 examples example NOUN bracis-28398 93 297 of of ADP bracis-28398 93 298 class class NOUN bracis-28398 93 299 \(y_i\ \(y_i\ NOUN bracis-28398 93 300 ) ) PUNCT bracis-28398 93 301 which which PRON bracis-28398 93 302 are be AUX bracis-28398 93 303 placed place VERB bracis-28398 93 304 in in ADP bracis-28398 93 305 a a DET bracis-28398 93 306 same same ADJ bracis-28398 93 307 disjunct disjunct NOUN bracis-28398 93 308 as as ADP bracis-28398 93 309 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-28398 93 310 ) ) PUNCT bracis-28398 93 311 , , PUNCT bracis-28398 93 312 where where SCONJ bracis-28398 93 313 disjuncts disjunct NOUN bracis-28398 93 314 are be AUX bracis-28398 93 315 defined define VERB bracis-28398 93 316 by by ADP bracis-28398 93 317 a a DET bracis-28398 93 318 decision decision NOUN bracis-28398 93 319 tree tree NOUN bracis-28398 93 320 algorithm algorithm NOUN bracis-28398 93 321 ; ; PUNCT bracis-28398 93 322 td td NOUN bracis-28398 93 323 ( ( PUNCT bracis-28398 93 324 tree tree NOUN bracis-28398 93 325 depth depth NOUN bracis-28398 93 326 ) ) PUNCT bracis-28398 93 327 gives give VERB bracis-28398 93 328 the the DET bracis-28398 93 329 depth depth NOUN bracis-28398 93 330 where where SCONJ bracis-28398 93 331 the the DET bracis-28398 93 332 instance instance NOUN bracis-28398 93 333 is be AUX bracis-28398 93 334 classified classify VERB bracis-28398 93 335 in in ADP bracis-28398 93 336 a a DET bracis-28398 93 337 decision decision NOUN bracis-28398 93 338 tree tree NOUN bracis-28398 93 339 model model NOUN bracis-28398 93 340 , , PUNCT bracis-28398 93 341 which which PRON bracis-28398 93 342 can can AUX bracis-28398 93 343 be be AUX bracis-28398 93 344 either either CCONJ bracis-28398 93 345 pruned prune VERB bracis-28398 93 346 ( ( PUNCT bracis-28398 93 347 \(td_p\ \(td_p\ PROPN bracis-28398 93 348 ) ) PUNCT bracis-28398 93 349 ) ) PUNCT bracis-28398 93 350 or or CCONJ bracis-28398 93 351 unpruned unpruned ADJ bracis-28398 93 352 ( ( PUNCT bracis-28398 93 353 \(td_u\ \(td_u\ NOUN bracis-28398 93 354 ) ) PUNCT bracis-28398 93 355 ) ) PUNCT bracis-28398 93 356 . . PUNCT bracis-28398 94 1 all all DET bracis-28398 94 2 measures measure NOUN bracis-28398 94 3 are be AUX bracis-28398 94 4 standardized standardize VERB bracis-28398 94 5 , , PUNCT bracis-28398 94 6 by by ADP bracis-28398 94 7 definition definition NOUN bracis-28398 94 8 , , PUNCT bracis-28398 94 9 in in ADP bracis-28398 94 10 the the DET bracis-28398 94 11 [ [ X bracis-28398 94 12 0 0 NUM bracis-28398 94 13 , , PUNCT bracis-28398 94 14 1 1 NUM bracis-28398 94 15 ] ] PUNCT bracis-28398 94 16 interval interval NOUN bracis-28398 94 17 , , PUNCT bracis-28398 94 18 so so SCONJ bracis-28398 94 19 that that SCONJ bracis-28398 94 20 larger large ADJ bracis-28398 94 21 values value NOUN bracis-28398 94 22 are be AUX bracis-28398 94 23 attributed attribute VERB bracis-28398 94 24 to to ADP bracis-28398 94 25 instances instance NOUN bracis-28398 94 26 hard hard ADJ bracis-28398 94 27 to to PART bracis-28398 94 28 classify classify VERB bracis-28398 94 29 according accord VERB bracis-28398 94 30 to to ADP bracis-28398 94 31 the the DET bracis-28398 94 32 measured measured ADJ bracis-28398 94 33 criterion criterion NOUN bracis-28398 94 34 . . PUNCT bracis-28398 95 1 3.2 3.2 NUM bracis-28398 95 2 hardness hardness NOUN bracis-28398 95 3 embedding embed VERB bracis-28398 95 4 the the DET bracis-28398 95 5 hardness hardness NOUN bracis-28398 95 6 embedding embedding NOUN bracis-28398 95 7 is be AUX bracis-28398 95 8 built build VERB bracis-28398 95 9 from from ADP bracis-28398 95 10 a a DET bracis-28398 95 11 composition composition NOUN bracis-28398 95 12 of of ADP bracis-28398 95 13 several several ADJ bracis-28398 95 14 sets set NOUN bracis-28398 95 15 . . PUNCT bracis-28398 96 1 all all DET bracis-28398 96 2 individual individual ADJ bracis-28398 96 3 instances instance NOUN bracis-28398 96 4 \(\textbf{x}_i \(\textbf{x}_i PROPN bracis-28398 96 5 \in \in PROPN bracis-28398 96 6 \mathcal \mathcal PROPN bracis-28398 96 7 d\ d\ PROPN bracis-28398 96 8 ) ) PUNCT bracis-28398 96 9 compose compose VERB bracis-28398 96 10 the the DET bracis-28398 96 11 instance instance NOUN bracis-28398 96 12 set set VERB bracis-28398 96 13 \(\mathcal \(\mathcal ADJ bracis-28398 96 14 i\ i\ NOUN bracis-28398 96 15 ) ) PUNCT bracis-28398 96 16 . . PUNCT bracis-28398 97 1 the the DET bracis-28398 97 2 algorithm algorithm NOUN bracis-28398 97 3 set set VERB bracis-28398 97 4 \(\mathcal \(\mathcal ADJ bracis-28398 97 5 a\ a\ NOUN bracis-28398 97 6 ) ) PUNCT bracis-28398 97 7 comprises comprise VERB bracis-28398 97 8 a a DET bracis-28398 97 9 portfolio portfolio NOUN bracis-28398 97 10 of of ADP bracis-28398 97 11 classification classification NOUN bracis-28398 97 12 algorithms algorithm NOUN bracis-28398 97 13 of of ADP bracis-28398 97 14 distinct distinct ADJ bracis-28398 97 15 biases bias NOUN bracis-28398 97 16 . . PUNCT bracis-28398 98 1 the the DET bracis-28398 98 2 performance performance NOUN bracis-28398 98 3 set set VERB bracis-28398 98 4 \(\mathcal \(\mathcal ADJ bracis-28398 98 5 p\ p\ NOUN bracis-28398 98 6 ) ) PUNCT bracis-28398 98 7 records record VERB bracis-28398 98 8 the the DET bracis-28398 98 9 predictive predictive ADJ bracis-28398 98 10 performance performance NOUN bracis-28398 98 11 obtained obtain VERB bracis-28398 98 12 by by ADP bracis-28398 98 13 each each DET bracis-28398 98 14 algorithm algorithm NOUN bracis-28398 98 15 in in ADP bracis-28398 98 16 \(\mathcal \(\mathcal ADJ bracis-28398 98 17 a\ a\ NOUN bracis-28398 98 18 ) ) PUNCT bracis-28398 98 19 for for ADP bracis-28398 98 20 every every DET bracis-28398 98 21 instance instance NOUN bracis-28398 98 22 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 98 23 ) ) PUNCT bracis-28398 98 24 . . PUNCT bracis-28398 99 1 the the DET bracis-28398 99 2 feature feature NOUN bracis-28398 99 3 set set VERB bracis-28398 99 4 \(\mathcal \(\mathcal ADJ bracis-28398 99 5 f\ f\ NOUN bracis-28398 99 6 ) ) PUNCT bracis-28398 99 7 contains contain VERB bracis-28398 99 8 the the DET bracis-28398 99 9 hm hm INTJ bracis-28398 99 10 extracted extract VERB bracis-28398 99 11 from from ADP bracis-28398 99 12 the the DET bracis-28398 99 13 instances instance NOUN bracis-28398 99 14 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-28398 99 15 ) ) PUNCT bracis-28398 99 16 . . PUNCT bracis-28398 100 1 therefore therefore ADV bracis-28398 100 2 , , PUNCT bracis-28398 100 3 for for ADP bracis-28398 100 4 each each DET bracis-28398 100 5 instance instance NOUN bracis-28398 100 6 \(\textbf{x}_i\in \(\textbf{x}_i\in PUNCT bracis-28398 100 7 \mathcal \mathcal PROPN bracis-28398 100 8 i\ i\ PROPN bracis-28398 100 9 ) ) PUNCT bracis-28398 100 10 and and CCONJ bracis-28398 100 11 for for ADP bracis-28398 100 12 each each DET bracis-28398 100 13 algorithm algorithm NOUN bracis-28398 100 14 \(\alpha \(\alpha VERB bracis-28398 100 15 _ _ PROPN bracis-28398 100 16 j j PROPN bracis-28398 100 17 \in \in PROPN bracis-28398 100 18 \mathcal \mathcal PROPN bracis-28398 100 19 a\ a\ PROPN bracis-28398 100 20 ) ) PUNCT bracis-28398 100 21 , , PUNCT bracis-28398 100 22 a a DET bracis-28398 100 23 feature feature NOUN bracis-28398 100 24 vector vector NOUN bracis-28398 100 25 \(\textbf{f}(\textbf{x}_i \(\textbf{f}(\textbf{x}_i ADJ bracis-28398 100 26 ) ) PUNCT bracis-28398 100 27 \in \in PROPN bracis-28398 100 28 \mathcal \mathcal PROPN bracis-28398 100 29 f\ f\ NOUN bracis-28398 100 30 ) ) PUNCT bracis-28398 100 31 , , PUNCT bracis-28398 100 32 and and CCONJ bracis-28398 100 33 an an DET bracis-28398 100 34 algorithm algorithm NOUN bracis-28398 100 35 performance performance NOUN bracis-28398 100 36 metric metric ADJ bracis-28398 100 37 \(p_m(\alpha \(p_m(\alpha X bracis-28398 101 1 _ _ PUNCT bracis-28398 101 2 j j X bracis-28398 101 3 , , PUNCT bracis-28398 101 4 \textbf{x}_i)\ \textbf{x}_i)\ ADV bracis-28398 101 5 ) ) PUNCT bracis-28398 101 6 are be AUX bracis-28398 101 7 measured measure VERB bracis-28398 101 8 . . PUNCT bracis-28398 102 1 the the DET bracis-28398 102 2 process process NOUN bracis-28398 102 3 is be AUX bracis-28398 102 4 repeated repeat VERB bracis-28398 102 5 for for ADP bracis-28398 102 6 all all DET bracis-28398 102 7 the the DET bracis-28398 102 8 instances instance NOUN bracis-28398 102 9 in in ADP bracis-28398 102 10 \(\mathcal \(\mathcal ADJ bracis-28398 102 11 i\ i\ NOUN bracis-28398 102 12 ) ) PUNCT bracis-28398 102 13 and and CCONJ bracis-28398 102 14 algorithms algorithm NOUN bracis-28398 102 15 in in ADP bracis-28398 102 16 \(\mathcal \(\mathcal ADJ bracis-28398 102 17 a\ a\ NOUN bracis-28398 102 18 ) ) PUNCT bracis-28398 102 19 , , PUNCT bracis-28398 102 20 generating generate VERB bracis-28398 102 21 a a DET bracis-28398 102 22 meta meta ADV bracis-28398 102 23 - - PUNCT bracis-28398 102 24 dataset dataset VERB bracis-28398 102 25 \(\mathcal \(\mathcal ADJ bracis-28398 102 26 m m NOUN bracis-28398 102 27 = = SYM bracis-28398 102 28 \{\mathcal \{\mathcal ADJ bracis-28398 102 29 i,\mathcal i,\mathcal ADJ bracis-28398 102 30 f,\mathcal f,\mathcal ADJ bracis-28398 102 31 a,\mathcal a,\mathcal ADJ bracis-28398 102 32 p\}\ p\}\ NOUN bracis-28398 102 33 ) ) PUNCT bracis-28398 102 34 . . PUNCT bracis-28398 103 1 the the DET bracis-28398 103 2 algorithms algorithm NOUN bracis-28398 103 3 included include VERB bracis-28398 103 4 in in ADP bracis-28398 103 5 the the DET bracis-28398 103 6 pool pool NOUN bracis-28398 103 7 \(\mathcal \(\mathcal ADJ bracis-28398 103 8 a\ a\ NOUN bracis-28398 103 9 ) ) PUNCT bracis-28398 103 10 are be AUX bracis-28398 103 11 : : PUNCT bracis-28398 103 12 bagging bagging NOUN bracis-28398 103 13 , , PUNCT bracis-28398 103 14 gradient gradient NOUN bracis-28398 103 15 boosting boosting NOUN bracis-28398 103 16 ( ( PUNCT bracis-28398 103 17 gb gb NOUN bracis-28398 103 18 ) ) PUNCT bracis-28398 103 19 , , PUNCT bracis-28398 103 20 support support NOUN bracis-28398 103 21 vector vector NOUN bracis-28398 103 22 machine machine NOUN bracis-28398 103 23 ( ( PUNCT bracis-28398 103 24 svm svm PROPN bracis-28398 103 25 , , PUNCT bracis-28398 103 26 with with ADP bracis-28398 103 27 both both CCONJ bracis-28398 103 28 linear linear ADJ bracis-28398 103 29 and and CCONJ bracis-28398 103 30 rbf rbf PROPN bracis-28398 103 31 kernels kernels PROPN bracis-28398 103 32 ) ) PUNCT bracis-28398 103 33 , , PUNCT bracis-28398 103 34 multilayer multilayer PROPN bracis-28398 103 35 perceptron perceptron PROPN bracis-28398 103 36 ( ( PUNCT bracis-28398 103 37 mlp mlp PROPN bracis-28398 103 38 ) ) PUNCT bracis-28398 103 39 , , PUNCT bracis-28398 103 40 logistic logistic ADJ bracis-28398 103 41 regression regression NOUN bracis-28398 103 42 ( ( PUNCT bracis-28398 103 43 lr lr NOUN bracis-28398 103 44 ) ) PUNCT bracis-28398 103 45 and and CCONJ bracis-28398 103 46 random random ADJ bracis-28398 103 47 forest forest NOUN bracis-28398 103 48 ( ( PUNCT bracis-28398 103 49 rf rf NOUN bracis-28398 103 50 ) ) PUNCT bracis-28398 103 51 . . PUNCT bracis-28398 104 1 they they PRON bracis-28398 104 2 present present VERB bracis-28398 104 3 distinct distinct ADJ bracis-28398 104 4 learning learning NOUN bracis-28398 104 5 mechanisms mechanism NOUN bracis-28398 104 6 and and CCONJ bracis-28398 104 7 are be AUX bracis-28398 104 8 commonly commonly ADV bracis-28398 104 9 adopted adopt VERB bracis-28398 104 10 to to PART bracis-28398 104 11 solve solve VERB bracis-28398 104 12 ml ml ADP bracis-28398 104 13 classification classification NOUN bracis-28398 104 14 tasks task NOUN bracis-28398 104 15 in in ADP bracis-28398 104 16 the the DET bracis-28398 104 17 literature literature NOUN bracis-28398 104 18 , , PUNCT bracis-28398 104 19 specially specially ADV bracis-28398 104 20 when when SCONJ bracis-28398 104 21 tabular tabular PROPN bracis-28398 104 22 datasets dataset NOUN bracis-28398 104 23 are be AUX bracis-28398 104 24 concerned concern VERB bracis-28398 104 25 . . PUNCT bracis-28398 105 1 to to PART bracis-28398 105 2 assess assess VERB bracis-28398 105 3 the the DET bracis-28398 105 4 performance performance NOUN bracis-28398 105 5 of of ADP bracis-28398 105 6 the the DET bracis-28398 105 7 algorithms algorithm NOUN bracis-28398 105 8 , , PUNCT bracis-28398 105 9 a a DET bracis-28398 105 10 five five NUM bracis-28398 105 11 - - ADJ bracis-28398 105 12 fold fold ADJ bracis-28398 105 13 cv cv ADJ bracis-28398 105 14 strategy strategy NOUN bracis-28398 105 15 is be AUX bracis-28398 105 16 used use VERB bracis-28398 105 17 . . PUNCT bracis-28398 106 1 their their PRON bracis-28398 106 2 performance performance NOUN bracis-28398 106 3 \(\mathcal \(\mathcal ADJ bracis-28398 106 4 p\ p\ NOUN bracis-28398 106 5 ) ) PUNCT bracis-28398 106 6 is be AUX bracis-28398 106 7 evaluated evaluate VERB bracis-28398 106 8 using use VERB bracis-28398 106 9 the the DET bracis-28398 106 10 log log NOUN bracis-28398 106 11 - - PUNCT bracis-28398 106 12 loss loss NOUN bracis-28398 106 13 error error NOUN bracis-28398 106 14 per per ADP bracis-28398 106 15 instance instance NOUN bracis-28398 106 16 . . PUNCT bracis-28398 107 1 the the DET bracis-28398 107 2 meta meta ADJ bracis-28398 107 3 - - PUNCT bracis-28398 107 4 dataset dataset VERB bracis-28398 107 5 \(\mathcal \(\mathcal ADJ bracis-28398 107 6 m\ m\ NOUN bracis-28398 107 7 ) ) PUNCT bracis-28398 107 8 is be AUX bracis-28398 107 9 next next ADJ bracis-28398 107 10 subject subject NOUN bracis-28398 107 11 to to ADP bracis-28398 107 12 a a DET bracis-28398 107 13 meta meta ADJ bracis-28398 107 14 - - PUNCT bracis-28398 107 15 feature feature NOUN bracis-28398 107 16 selection selection NOUN bracis-28398 107 17 step step NOUN bracis-28398 107 18 so so SCONJ bracis-28398 107 19 that that SCONJ bracis-28398 107 20 only only ADV bracis-28398 107 21 the the DET bracis-28398 107 22 most most ADV bracis-28398 107 23 informative informative ADJ bracis-28398 107 24 meta meta NOUN bracis-28398 107 25 - - PUNCT bracis-28398 107 26 features feature NOUN bracis-28398 107 27 are be AUX bracis-28398 107 28 preserved preserve VERB bracis-28398 107 29 . . PUNCT bracis-28398 108 1 this this DET bracis-28398 108 2 selection selection NOUN bracis-28398 108 3 is be AUX bracis-28398 108 4 performed perform VERB bracis-28398 108 5 using use VERB bracis-28398 108 6 a a DET bracis-28398 108 7 neighborhood neighborhood NOUN bracis-28398 108 8 component component NOUN bracis-28398 108 9 feature feature NOUN bracis-28398 108 10 selection selection NOUN bracis-28398 108 11 ( ( PUNCT bracis-28398 108 12 ncfs ncfs NOUN bracis-28398 108 13 ) ) PUNCT bracis-28398 108 14 algorithm algorithm NOUN bracis-28398 108 15     SPACE bracis-28398 109 1 [ [ X bracis-28398 109 2 21 21 NUM bracis-28398 109 3 ] ] PUNCT bracis-28398 109 4 , , PUNCT bracis-28398 109 5 where where SCONJ bracis-28398 109 6 we we PRON bracis-28398 109 7 seek seek VERB bracis-28398 109 8 to to PART bracis-28398 109 9 find find VERB bracis-28398 109 10 meta meta NOUN bracis-28398 109 11 - - PUNCT bracis-28398 109 12 features feature VERB bracis-28398 109 13 more more ADV bracis-28398 109 14 related related ADJ bracis-28398 109 15 to to ADP bracis-28398 109 16 the the DET bracis-28398 109 17 predictive predictive ADJ bracis-28398 109 18 performance performance NOUN bracis-28398 109 19 of of ADP bracis-28398 109 20 the the DET bracis-28398 109 21 classifiers classifier NOUN bracis-28398 109 22 in in ADP bracis-28398 109 23 \(\mathcal \(\mathcal ADJ bracis-28398 109 24 a\ a\ NOUN bracis-28398 109 25 ) ) PUNCT bracis-28398 109 26 . . PUNCT bracis-28398 110 1 the the DET bracis-28398 110 2 resulting result VERB bracis-28398 110 3 meta meta ADV bracis-28398 110 4 - - PUNCT bracis-28398 110 5 dataset dataset NOUN bracis-28398 110 6 is be AUX bracis-28398 110 7 then then ADV bracis-28398 110 8 projected project VERB bracis-28398 110 9 into into ADP bracis-28398 110 10 a a DET bracis-28398 110 11 2 2 NUM bracis-28398 110 12 - - PUNCT bracis-28398 110 13 d d NOUN bracis-28398 110 14 embedding embed VERB bracis-28398 110 15 presenting present VERB bracis-28398 110 16 linear linear ADJ bracis-28398 110 17 trends trend NOUN bracis-28398 110 18 in in ADP bracis-28398 110 19 the the DET bracis-28398 110 20 meta meta NOUN bracis-28398 110 21 - - PUNCT bracis-28398 110 22 features feature NOUN bracis-28398 110 23 and and CCONJ bracis-28398 110 24 algorithmic algorithmic ADJ bracis-28398 110 25 performance performance NOUN bracis-28398 110 26 measures measure NOUN bracis-28398 110 27 , , PUNCT bracis-28398 110 28 also also ADV bracis-28398 110 29 named name VERB bracis-28398 110 30 instance instance NOUN bracis-28398 110 31 space space NOUN bracis-28398 110 32 ( ( PUNCT bracis-28398 110 33 is be AUX bracis-28398 110 34 ) ) PUNCT bracis-28398 110 35 , , PUNCT bracis-28398 110 36 as as SCONJ bracis-28398 110 37 described describe VERB bracis-28398 110 38 in in ADP bracis-28398 110 39     SPACE bracis-28398 110 40 [ [ X bracis-28398 110 41 8 8 NUM bracis-28398 110 42 , , PUNCT bracis-28398 110 43 11 11 NUM bracis-28398 110 44 ] ] PUNCT bracis-28398 110 45 . . PUNCT bracis-28398 111 1 the the DET bracis-28398 111 2 result result NOUN bracis-28398 111 3 is be AUX bracis-28398 111 4 a a DET bracis-28398 111 5 projection projection NOUN bracis-28398 111 6 matrix matrix NOUN bracis-28398 111 7 able able ADJ bracis-28398 111 8 to to PART bracis-28398 111 9 map map VERB bracis-28398 111 10 each each DET bracis-28398 111 11 instance instance NOUN bracis-28398 111 12 to to ADP bracis-28398 111 13 the the DET bracis-28398 111 14 2 2 NUM bracis-28398 111 15 - - PUNCT bracis-28398 111 16 d d NOUN bracis-28398 111 17 hardness hardness NOUN bracis-28398 111 18 embedding embedding NOUN bracis-28398 111 19 based base VERB bracis-28398 111 20 on on ADP bracis-28398 111 21 the the DET bracis-28398 111 22 selected select VERB bracis-28398 111 23 meta meta ADJ bracis-28398 111 24 - - PUNCT bracis-28398 111 25 features feature NOUN bracis-28398 111 26 values value NOUN bracis-28398 111 27 . . PUNCT bracis-28398 112 1 a a DET bracis-28398 112 2 further further ADJ bracis-28398 112 3 rotation rotation NOUN bracis-28398 112 4 step step NOUN bracis-28398 112 5 is be AUX bracis-28398 112 6 also also ADV bracis-28398 112 7 included include VERB bracis-28398 112 8 , , PUNCT bracis-28398 112 9 so so SCONJ bracis-28398 112 10 that that SCONJ bracis-28398 112 11 the the DET bracis-28398 112 12 hard hard ADJ bracis-28398 112 13 instances instance NOUN bracis-28398 112 14 are be AUX bracis-28398 112 15 placed place VERB bracis-28398 112 16 in in ADP bracis-28398 112 17 the the DET bracis-28398 112 18 upper upper ADJ bracis-28398 112 19 left left NOUN bracis-28398 112 20 quadrant quadrant NOUN bracis-28398 112 21 and and CCONJ bracis-28398 112 22 the the DET bracis-28398 112 23 easier easy ADJ bracis-28398 112 24 instances instance NOUN bracis-28398 112 25 are be AUX bracis-28398 112 26 in in ADP bracis-28398 112 27 the the DET bracis-28398 112 28 lower low ADJ bracis-28398 112 29 right right ADJ bracis-28398 112 30 quadrant quadrant NOUN bracis-28398 112 31 of of ADP bracis-28398 112 32 the the DET bracis-28398 112 33 hardness hardness NOUN bracis-28398 112 34 embedding embed VERB bracis-28398 112 35 . . PUNCT bracis-28398 113 1 figure figure NOUN bracis-28398 113 2     SPACE bracis-28398 113 3 2 2 NUM bracis-28398 113 4 presents present VERB bracis-28398 113 5 examples example NOUN bracis-28398 113 6 of of ADP bracis-28398 113 7 hardness hardness ADJ bracis-28398 113 8 embeddings embedding NOUN bracis-28398 113 9 of of ADP bracis-28398 113 10 three three NUM bracis-28398 113 11 datasets dataset NOUN bracis-28398 113 12 for for ADP bracis-28398 113 13 predicting predict VERB bracis-28398 113 14 the the DET bracis-28398 113 15 outcome outcome NOUN bracis-28398 113 16 of of ADP bracis-28398 113 17 covid covid PROPN bracis-28398 113 18 patient patient ADJ bracis-28398 113 19 severity severity NOUN bracis-28398 113 20 . . PUNCT bracis-28398 114 1 the the DET bracis-28398 114 2 first first ADJ bracis-28398 114 3 and and CCONJ bracis-28398 114 4 third third ADJ bracis-28398 114 5 datasets dataset NOUN bracis-28398 114 6 contain contain VERB bracis-28398 114 7 the the DET bracis-28398 114 8 symptoms symptom NOUN bracis-28398 114 9 and and CCONJ bracis-28398 114 10 comorbidities comorbiditie NOUN bracis-28398 114 11 of of ADP bracis-28398 114 12 citizens citizen NOUN bracis-28398 114 13 undergoing undergo VERB bracis-28398 114 14 diagnosis diagnosis NOUN bracis-28398 114 15 tests test NOUN bracis-28398 114 16 . . PUNCT bracis-28398 115 1 in in ADP bracis-28398 115 2 the the DET bracis-28398 115 3 second second ADJ bracis-28398 115 4 example example NOUN bracis-28398 115 5 , , PUNCT bracis-28398 115 6 patients patient NOUN bracis-28398 115 7 of of ADP bracis-28398 115 8 an an DET bracis-28398 115 9 hospital hospital NOUN bracis-28398 115 10 from from ADP bracis-28398 115 11 the the DET bracis-28398 115 12 são são PROPN bracis-28398 115 13 paulo paulo PROPN bracis-28398 115 14 metropolitan metropolitan PROPN bracis-28398 115 15 area area PROPN bracis-28398 115 16 ( ( PUNCT bracis-28398 115 17 brazil brazil PROPN bracis-28398 115 18 ) ) PUNCT bracis-28398 115 19 are be AUX bracis-28398 115 20 prognosed prognose VERB bracis-28398 115 21 based base VERB bracis-28398 115 22 on on ADP bracis-28398 115 23 a a DET bracis-28398 115 24 set set NOUN bracis-28398 115 25 of of ADP bracis-28398 115 26 routine routine ADJ bracis-28398 115 27 laboratory laboratory NOUN bracis-28398 115 28 tests test NOUN bracis-28398 115 29 . . PUNCT bracis-28398 116 1 in in ADP bracis-28398 116 2 all all DET bracis-28398 116 3 cases case NOUN bracis-28398 116 4 , , PUNCT bracis-28398 116 5 classes class NOUN bracis-28398 116 6 are be AUX bracis-28398 116 7 represented represent VERB bracis-28398 116 8 by by ADP bracis-28398 116 9 different different ADJ bracis-28398 116 10 shapes shape NOUN bracis-28398 116 11 . . PUNCT bracis-28398 117 1 the the DET bracis-28398 117 2 colors color NOUN bracis-28398 117 3 give give VERB bracis-28398 117 4 the the DET bracis-28398 117 5 instance instance NOUN bracis-28398 117 6 hardness hardness NOUN bracis-28398 117 7 level level NOUN bracis-28398 117 8 of of ADP bracis-28398 117 9 the the DET bracis-28398 117 10 instances instance NOUN bracis-28398 117 11 as as SCONJ bracis-28398 117 12 computed compute VERB bracis-28398 117 13 by by ADP bracis-28398 117 14 eq eq PROPN bracis-28398 117 15 . . PUNCT bracis-28398 117 16     SPACE bracis-28398 117 17 1 1 NUM bracis-28398 117 18 . . PUNCT bracis-28398 118 1 the the PRON bracis-28398 118 2 higher high ADJ bracis-28398 118 3 the the DET bracis-28398 118 4 ih ih NOUN bracis-28398 118 5 value value NOUN bracis-28398 118 6 , , PUNCT bracis-28398 118 7 the the DET bracis-28398 118 8 redder red ADJ bracis-28398 118 9 the the DET bracis-28398 118 10 color color NOUN bracis-28398 118 11 ; ; PUNCT bracis-28398 118 12 the the PRON bracis-28398 118 13 lower low ADJ bracis-28398 118 14 the the DET bracis-28398 118 15 ih ih NOUN bracis-28398 118 16 value value NOUN bracis-28398 118 17 , , PUNCT bracis-28398 118 18 the the DET bracis-28398 118 19 bluer bluer ADJ bracis-28398 118 20 the the DET bracis-28398 118 21 color color NOUN bracis-28398 118 22 . . PUNCT bracis-28398 119 1 there there PRON bracis-28398 119 2 is be VERB bracis-28398 119 3 a a DET bracis-28398 119 4 mix mix NOUN bracis-28398 119 5 of of ADP bracis-28398 119 6 patients patient NOUN bracis-28398 119 7 that that PRON bracis-28398 119 8 are be AUX bracis-28398 119 9 easy easy ADJ bracis-28398 119 10 or or CCONJ bracis-28398 119 11 hard hard ADJ bracis-28398 119 12 to to PART bracis-28398 119 13 classify classify VERB bracis-28398 119 14 per per ADP bracis-28398 119 15 class class NOUN bracis-28398 119 16 , , PUNCT bracis-28398 119 17 with with ADP bracis-28398 119 18 both both DET bracis-28398 119 19 triangles triangle NOUN bracis-28398 119 20 and and CCONJ bracis-28398 119 21 circles circle NOUN bracis-28398 119 22 in in ADP bracis-28398 119 23 the the DET bracis-28398 119 24 upper upper ADJ bracis-28398 119 25 left left NOUN bracis-28398 119 26 or or CCONJ bracis-28398 119 27 lower low ADJ bracis-28398 119 28 right right ADJ bracis-28398 119 29 regions region NOUN bracis-28398 119 30 of of ADP bracis-28398 119 31 the the DET bracis-28398 119 32 plot plot NOUN bracis-28398 119 33 . . PUNCT bracis-28398 120 1 fig fig NOUN bracis-28398 120 2 . . PUNCT bracis-28398 121 1 2 2 X bracis-28398 121 2 . . X bracis-28398 121 3 hardness hardness NOUN bracis-28398 121 4 embedding embedding NOUN bracis-28398 121 5 of of ADP bracis-28398 121 6 three three NUM bracis-28398 121 7 datasets dataset NOUN bracis-28398 121 8 with with ADP bracis-28398 121 9 contrasting contrast VERB bracis-28398 121 10 profiles profile NOUN bracis-28398 121 11 : : PUNCT bracis-28398 121 12 ( ( PUNCT bracis-28398 121 13 a a X bracis-28398 121 14 ) ) PUNCT bracis-28398 121 15 severity severity NOUN bracis-28398 121 16 ; ; PUNCT bracis-28398 121 17 ( ( PUNCT bracis-28398 121 18 b b X bracis-28398 121 19 ) ) PUNCT bracis-28398 121 20 hospital hospital NOUN bracis-28398 121 21 1 1 NUM bracis-28398 121 22 ; ; PUNCT bracis-28398 121 23 ( ( PUNCT bracis-28398 121 24 c c X bracis-28398 121 25 ) ) PUNCT bracis-28398 121 26 hospitalization hospitalization NOUN bracis-28398 121 27 . . PUNCT bracis-28398 122 1 full full ADJ bracis-28398 122 2 size size NOUN bracis-28398 122 3 image image NOUN bracis-28398 122 4 whilst whilst SCONJ bracis-28398 122 5 this this DET bracis-28398 122 6 plot plot NOUN bracis-28398 122 7 can can AUX bracis-28398 122 8 be be AUX bracis-28398 122 9 colored color VERB bracis-28398 122 10 according accord VERB bracis-28398 122 11 to to ADP bracis-28398 122 12 all all DET bracis-28398 122 13 meta meta ADJ bracis-28398 122 14 - - PUNCT bracis-28398 122 15 features feature VERB bracis-28398 122 16 \(\mathcal \(\mathcal ADJ bracis-28398 122 17 f\ f\ NOUN bracis-28398 122 18 ) ) PUNCT bracis-28398 122 19 values value NOUN bracis-28398 122 20 , , PUNCT bracis-28398 122 21 performance performance NOUN bracis-28398 122 22 measures measure NOUN bracis-28398 122 23 \(\mathcal \(\mathcal ADJ bracis-28398 122 24 p\ p\ NOUN bracis-28398 122 25 ) ) PUNCT bracis-28398 122 26 values value NOUN bracis-28398 122 27 of of ADP bracis-28398 122 28 the the DET bracis-28398 122 29 classification classification NOUN bracis-28398 122 30 algorithms algorithms NOUN bracis-28398 122 31 \(\mathcal \(\mathcal NOUN bracis-28398 122 32 a\ a\ NOUN bracis-28398 122 33 ) ) PUNCT bracis-28398 122 34 and and CCONJ bracis-28398 122 35 even even ADV bracis-28398 122 36 by by ADP bracis-28398 122 37 the the DET bracis-28398 122 38 original original ADJ bracis-28398 122 39 input input NOUN bracis-28398 122 40 features feature VERB bracis-28398 122 41 values value NOUN bracis-28398 122 42 of of ADP bracis-28398 122 43 the the DET bracis-28398 122 44 dataset dataset ADJ bracis-28398 122 45 \(\mathcal \(\mathcal ADJ bracis-28398 122 46 d\ d\ NOUN bracis-28398 122 47 ) ) PUNCT bracis-28398 122 48 , , PUNCT bracis-28398 122 49 selections selection NOUN bracis-28398 122 50 of of ADP bracis-28398 122 51 specific specific ADJ bracis-28398 122 52 regions region NOUN bracis-28398 122 53 can can AUX bracis-28398 122 54 also also ADV bracis-28398 122 55 be be AUX bracis-28398 122 56 saved save VERB bracis-28398 122 57 and and CCONJ bracis-28398 122 58 scrutinized scrutinize VERB bracis-28398 122 59 . . PUNCT bracis-28398 123 1 3.3 3.3 NUM bracis-28398 123 2 instance instance NOUN bracis-28398 123 3 easiness easiness NOUN bracis-28398 123 4 and and CCONJ bracis-28398 123 5     SPACE bracis-28398 123 6 instance instance NOUN bracis-28398 123 7 hardness hardness NOUN bracis-28398 123 8 footprints footprint NOUN bracis-28398 123 9 a a DET bracis-28398 123 10 footprint footprint NOUN bracis-28398 123 11 is be AUX bracis-28398 123 12 defined define VERB bracis-28398 123 13 in in ADP bracis-28398 123 14     SPACE bracis-28398 123 15 [ [ X bracis-28398 123 16 22 22 NUM bracis-28398 123 17 , , PUNCT bracis-28398 123 18 23 23 NUM bracis-28398 123 19 ] ] PUNCT bracis-28398 123 20 as as ADP bracis-28398 123 21 a a DET bracis-28398 123 22 region region NOUN bracis-28398 123 23 in in ADP bracis-28398 123 24 the the DET bracis-28398 123 25 instance instance NOUN bracis-28398 123 26 space space NOUN bracis-28398 123 27 where where SCONJ bracis-28398 123 28 an an DET bracis-28398 123 29 algorithm algorithm NOUN bracis-28398 123 30 is be AUX bracis-28398 123 31 expected expect VERB bracis-28398 123 32 to to PART bracis-28398 123 33 perform perform VERB bracis-28398 123 34 well well ADV bracis-28398 123 35 based base VERB bracis-28398 123 36 on on ADP bracis-28398 123 37 inference inference NOUN bracis-28398 123 38 from from ADP bracis-28398 123 39 empirical empirical ADJ bracis-28398 123 40 performance performance NOUN bracis-28398 123 41 analysis analysis NOUN bracis-28398 123 42 . . PUNCT bracis-28398 124 1 here here ADV bracis-28398 124 2 we we PRON bracis-28398 124 3 use use VERB bracis-28398 124 4 such such DET bracis-28398 124 5 a a DET bracis-28398 124 6 concept concept NOUN bracis-28398 124 7 in in ADP bracis-28398 124 8 order order NOUN bracis-28398 124 9 to to PART bracis-28398 124 10 define define VERB bracis-28398 124 11 regions region NOUN bracis-28398 124 12 from from ADP bracis-28398 124 13 the the DET bracis-28398 124 14 hardness hardness NOUN bracis-28398 124 15 embedding embedding NOUN bracis-28398 124 16 which which PRON bracis-28398 124 17 concentrate concentrate VERB bracis-28398 124 18 more more ADV bracis-28398 124 19 easy easy ADJ bracis-28398 124 20 or or CCONJ bracis-28398 124 21 hard hard ADJ bracis-28398 124 22 instances instance NOUN bracis-28398 124 23 . . PUNCT bracis-28398 125 1 they they PRON bracis-28398 125 2 are be AUX bracis-28398 125 3 named name VERB bracis-28398 125 4 easiness easiness NOUN bracis-28398 125 5 and and CCONJ bracis-28398 125 6 hardness hardness NOUN bracis-28398 125 7 footprints footprint NOUN bracis-28398 125 8 , , PUNCT bracis-28398 125 9 respectively respectively ADV bracis-28398 125 10 . . PUNCT bracis-28398 126 1 in in ADP bracis-28398 126 2 order order NOUN bracis-28398 126 3 to to PART bracis-28398 126 4 construct construct VERB bracis-28398 126 5 these these DET bracis-28398 126 6 footprints footprint NOUN bracis-28398 126 7 , , PUNCT bracis-28398 126 8 first first ADV bracis-28398 126 9 the the DET bracis-28398 126 10 instances instance NOUN bracis-28398 126 11 must must AUX bracis-28398 126 12 be be AUX bracis-28398 126 13 categorized categorize VERB bracis-28398 126 14 as as ADP bracis-28398 126 15 either either CCONJ bracis-28398 126 16 easy easy ADJ bracis-28398 126 17 or or CCONJ bracis-28398 126 18 hard hard ADJ bracis-28398 126 19 . . PUNCT bracis-28398 127 1 this this PRON bracis-28398 127 2 can can AUX bracis-28398 127 3 be be AUX bracis-28398 127 4 done do VERB bracis-28398 127 5 by by ADP bracis-28398 127 6 imposing impose VERB bracis-28398 127 7 different different ADJ bracis-28398 127 8 thresholds threshold NOUN bracis-28398 127 9 on on ADP bracis-28398 127 10 ih ih NOUN bracis-28398 127 11 values value NOUN bracis-28398 127 12 . . PUNCT bracis-28398 128 1 in in ADP bracis-28398 128 2 fact fact NOUN bracis-28398 128 3 , , PUNCT bracis-28398 128 4 we we PRON bracis-28398 128 5 consider consider VERB bracis-28398 128 6 three three NUM bracis-28398 128 7 categories category NOUN bracis-28398 128 8 : : PUNCT bracis-28398 128 9 easy easy ADJ bracis-28398 128 10 , , PUNCT bracis-28398 128 11 hard hard ADJ bracis-28398 128 12 and and CCONJ bracis-28398 128 13 others other NOUN bracis-28398 128 14 . . PUNCT bracis-28398 129 1 the the DET bracis-28398 129 2 later later ADJ bracis-28398 129 3 category category NOUN bracis-28398 129 4 encompasses encompass VERB bracis-28398 129 5 instances instance NOUN bracis-28398 129 6 that that PRON bracis-28398 129 7 have have VERB bracis-28398 129 8 intermediate intermediate ADJ bracis-28398 129 9 hardness hardness ADJ bracis-28398 129 10 levels level NOUN bracis-28398 129 11 . . PUNCT bracis-28398 130 1 we we PRON bracis-28398 130 2 focus focus VERB bracis-28398 130 3 our our PRON bracis-28398 130 4 analysis analysis NOUN bracis-28398 130 5 on on ADP bracis-28398 130 6 contrasting contrast VERB bracis-28398 130 7 easy easy ADJ bracis-28398 130 8 and and CCONJ bracis-28398 130 9 hard hard ADJ bracis-28398 130 10 instances instance NOUN bracis-28398 130 11 . . PUNCT bracis-28398 131 1 ultimately ultimately ADV bracis-28398 131 2 , , PUNCT bracis-28398 131 3 the the DET bracis-28398 131 4 objective objective NOUN bracis-28398 131 5 is be AUX bracis-28398 131 6 to to PART bracis-28398 131 7 understand understand VERB bracis-28398 131 8 why why SCONJ bracis-28398 131 9 some some DET bracis-28398 131 10 instances instance NOUN bracis-28398 131 11 are be AUX bracis-28398 131 12 hard hard ADJ bracis-28398 131 13 to to PART bracis-28398 131 14 classify classify VERB bracis-28398 131 15 . . PUNCT bracis-28398 132 1 for for ADP bracis-28398 132 2 segmenting segment VERB bracis-28398 132 3 the the DET bracis-28398 132 4 instances instance NOUN bracis-28398 132 5 , , PUNCT bracis-28398 132 6 we we PRON bracis-28398 132 7 consider consider VERB bracis-28398 132 8 a a DET bracis-28398 132 9 threshold threshold NOUN bracis-28398 132 10 t t PROPN bracis-28398 132 11 , , PUNCT bracis-28398 132 12 which which PRON bracis-28398 132 13 must must AUX bracis-28398 132 14 be be AUX bracis-28398 132 15 higher high ADJ bracis-28398 132 16 than than ADP bracis-28398 132 17 0 0 NUM bracis-28398 132 18 and and CCONJ bracis-28398 132 19 lower low ADJ bracis-28398 132 20 than than ADP bracis-28398 132 21 0.5 0.5 NUM bracis-28398 132 22 . . PUNCT bracis-28398 133 1 since since SCONJ bracis-28398 133 2 the the DET bracis-28398 133 3 ih ih NOUN bracis-28398 133 4 values value NOUN bracis-28398 133 5 are be AUX bracis-28398 133 6 bounded bound VERB bracis-28398 133 7 in in ADP bracis-28398 133 8 the the DET bracis-28398 133 9 [ [ X bracis-28398 133 10 0 0 NUM bracis-28398 133 11 , , PUNCT bracis-28398 133 12 1 1 NUM bracis-28398 133 13 ] ] PUNCT bracis-28398 133 14 interval interval NOUN bracis-28398 133 15 , , PUNCT bracis-28398 133 16 we we PRON bracis-28398 133 17 have have AUX bracis-28398 133 18 : : PUNCT bracis-28398 133 19 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 133 20 ) ) PUNCT bracis-28398 133 21 is be AUX bracis-28398 133 22 considered consider VERB bracis-28398 133 23 easy easy ADJ bracis-28398 133 24 if if SCONJ bracis-28398 133 25 \(ih_{\mathcal \(ih_{\mathcal PROPN bracis-28398 133 26 a}(\textbf{x}_i a}(\textbf{x}_i PROPN bracis-28398 133 27 , , PUNCT bracis-28398 133 28 y_i y_i NOUN bracis-28398 133 29 ) ) PUNCT bracis-28398 133 30 \le \le PROPN bracis-28398 133 31 t\ t\ PROPN bracis-28398 133 32 ) ) PUNCT bracis-28398 133 33 ; ; PUNCT bracis-28398 133 34 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-28398 133 35 ) ) PUNCT bracis-28398 133 36 is be AUX bracis-28398 133 37 considered consider VERB bracis-28398 133 38 hard hard ADJ bracis-28398 133 39 if if SCONJ bracis-28398 133 40 \(ih_{\mathcal \(ih_{\mathcal PROPN bracis-28398 133 41 a}(\textbf{x}_i a}(\textbf{x}_i PROPN bracis-28398 133 42 , , PUNCT bracis-28398 133 43 y_i y_i NOUN bracis-28398 133 44 ) ) PUNCT bracis-28398 133 45 \ge \ge PROPN bracis-28398 133 46 1 1 NUM bracis-28398 133 47 - - PUNCT bracis-28398 133 48 t\ t\ NOUN bracis-28398 133 49 ) ) PUNCT bracis-28398 133 50 . . PUNCT bracis-28398 134 1 then then ADV bracis-28398 134 2 for for ADP bracis-28398 134 3 each each DET bracis-28398 134 4 set set NOUN bracis-28398 134 5 of of ADP bracis-28398 134 6 instances instance NOUN bracis-28398 134 7 , , PUNCT bracis-28398 134 8 easy easy ADJ bracis-28398 134 9 or or CCONJ bracis-28398 134 10 hard hard ADJ bracis-28398 134 11 , , PUNCT bracis-28398 134 12 the the DET bracis-28398 134 13 dbscan dbscan PROPN bracis-28398 134 14 algorithm algorithm PROPN bracis-28398 134 15     SPACE bracis-28398 135 1 [ [ X bracis-28398 135 2 24 24 NUM bracis-28398 135 3 ] ] PUNCT bracis-28398 135 4 is be AUX bracis-28398 135 5 used use VERB bracis-28398 135 6 to to PART bracis-28398 135 7 identify identify VERB bracis-28398 135 8 high high ADJ bracis-28398 135 9 density density NOUN bracis-28398 135 10 clusters cluster NOUN bracis-28398 135 11 in in ADP bracis-28398 135 12 the the DET bracis-28398 135 13 instance instance NOUN bracis-28398 135 14 ( ( PUNCT bracis-28398 135 15 hardness hardness NOUN bracis-28398 135 16 ) ) PUNCT bracis-28398 135 17 space space NOUN bracis-28398 135 18 . . PUNCT bracis-28398 136 1 \(\alpha \(\alpha PROPN bracis-28398 136 2 \)-shapes \)-shape NOUN bracis-28398 136 3 are be AUX bracis-28398 136 4 used use VERB bracis-28398 136 5 to to PART bracis-28398 136 6 construct construct VERB bracis-28398 136 7 hulls hull NOUN bracis-28398 136 8 which which PRON bracis-28398 136 9 enclose enclose VERB bracis-28398 136 10 all all DET bracis-28398 136 11 the the DET bracis-28398 136 12 points point NOUN bracis-28398 136 13 within within ADP bracis-28398 136 14 the the DET bracis-28398 136 15 clusters cluster NOUN bracis-28398 136 16     SPACE bracis-28398 137 1 [ [ X bracis-28398 137 2 25 25 NUM bracis-28398 137 3 ] ] PUNCT bracis-28398 137 4 . . PUNCT bracis-28398 138 1 for for ADP bracis-28398 138 2 each each DET bracis-28398 138 3 cluster cluster NOUN bracis-28398 138 4 hull hull NOUN bracis-28398 138 5 , , PUNCT bracis-28398 138 6 a a DET bracis-28398 138 7 delaunay delaunay PROPN bracis-28398 138 8 triangulation triangulation NOUN bracis-28398 138 9 creates create VERB bracis-28398 138 10 a a DET bracis-28398 138 11 partition partition NOUN bracis-28398 138 12 , , PUNCT bracis-28398 138 13 and and CCONJ bracis-28398 138 14 those those DET bracis-28398 138 15 triangles triangle NOUN bracis-28398 138 16 that that PRON bracis-28398 138 17 do do AUX bracis-28398 138 18 not not PART bracis-28398 138 19 satisfy satisfy VERB bracis-28398 138 20 a a DET bracis-28398 138 21 minimum minimum ADJ bracis-28398 138 22 purity purity NOUN bracis-28398 138 23 ( ( PUNCT bracis-28398 138 24 defined define VERB bracis-28398 138 25 as as ADP bracis-28398 138 26 the the DET bracis-28398 138 27 percentage percentage NOUN bracis-28398 138 28 of of ADP bracis-28398 138 29 easy easy ADJ bracis-28398 138 30 or or CCONJ bracis-28398 138 31 hard hard ADJ bracis-28398 138 32 instances instance NOUN bracis-28398 138 33 enclosed enclose VERB bracis-28398 138 34 within within ADP bracis-28398 138 35 it it PRON bracis-28398 138 36 , , PUNCT bracis-28398 138 37 which which PRON bracis-28398 138 38 is be AUX bracis-28398 138 39 0.7 0.7 NUM bracis-28398 138 40 by by ADP bracis-28398 138 41 default default NOUN bracis-28398 138 42 ) ) PUNCT bracis-28398 138 43 requirement requirement NOUN bracis-28398 138 44 are be AUX bracis-28398 138 45 removed remove VERB bracis-28398 138 46 . . PUNCT bracis-28398 139 1 the the DET bracis-28398 139 2 union union NOUN bracis-28398 139 3 of of ADP bracis-28398 139 4 the the DET bracis-28398 139 5 remaining remain VERB bracis-28398 139 6 triangles triangle NOUN bracis-28398 139 7 gives give VERB bracis-28398 139 8 the the DET bracis-28398 139 9 corresponding corresponding ADJ bracis-28398 139 10 easiness easiness NOUN bracis-28398 139 11 or or CCONJ bracis-28398 139 12 hardness hardness NOUN bracis-28398 139 13 footprint footprint NOUN bracis-28398 139 14 . . PUNCT bracis-28398 140 1 as as ADP bracis-28398 140 2 in in ADP bracis-28398 140 3     SPACE bracis-28398 140 4 [ [ X bracis-28398 140 5 22 22 NUM bracis-28398 140 6 ] ] PUNCT bracis-28398 140 7 , , PUNCT bracis-28398 140 8 it it PRON bracis-28398 140 9 is be AUX bracis-28398 140 10 also also ADV bracis-28398 140 11 possible possible ADJ bracis-28398 140 12 to to PART bracis-28398 140 13 extract extract VERB bracis-28398 140 14 the the DET bracis-28398 140 15 following follow VERB bracis-28398 140 16 objective objective ADJ bracis-28398 140 17 measures measure NOUN bracis-28398 140 18 from from ADP bracis-28398 140 19 the the DET bracis-28398 140 20 footprints footprint NOUN bracis-28398 140 21 : : PUNCT bracis-28398 141 1 1 1 X bracis-28398 141 2 . . X bracis-28398 141 3 the the DET bracis-28398 141 4 area area NOUN bracis-28398 141 5 of of ADP bracis-28398 141 6 the the DET bracis-28398 141 7 footprint footprint NOUN bracis-28398 141 8 in in ADP bracis-28398 141 9 the the DET bracis-28398 141 10 2 2 NUM bracis-28398 141 11 - - PUNCT bracis-28398 141 12 d d NOUN bracis-28398 141 13 hardness hardness NOUN bracis-28398 141 14 embedding embedding NOUN bracis-28398 141 15 , , PUNCT bracis-28398 141 16 normalized normalize VERB bracis-28398 141 17 by by ADP bracis-28398 141 18 the the DET bracis-28398 141 19 total total ADJ bracis-28398 141 20 area area NOUN bracis-28398 141 21 of of ADP bracis-28398 141 22 the the DET bracis-28398 141 23 space space NOUN bracis-28398 141 24 ; ; PUNCT bracis-28398 141 25 2 2 X bracis-28398 141 26 . . X bracis-28398 141 27 the the DET bracis-28398 141 28 density density NOUN bracis-28398 141 29 of of ADP bracis-28398 141 30 the the DET bracis-28398 141 31 footprint footprint NOUN bracis-28398 141 32 , , PUNCT bracis-28398 141 33 computed compute VERB bracis-28398 141 34 as as ADP bracis-28398 141 35 the the DET bracis-28398 141 36 ratio ratio NOUN bracis-28398 141 37 between between ADP bracis-28398 141 38 the the DET bracis-28398 141 39 number number NOUN bracis-28398 141 40 of of ADP bracis-28398 141 41 instances instance NOUN bracis-28398 141 42 enclosed enclose VERB bracis-28398 141 43 by by ADP bracis-28398 141 44 the the DET bracis-28398 141 45 footprint footprint NOUN bracis-28398 141 46 and and CCONJ bracis-28398 141 47 its its PRON bracis-28398 141 48 area area NOUN bracis-28398 141 49 ; ; PUNCT bracis-28398 141 50 3 3 X bracis-28398 141 51 . . X bracis-28398 141 52 the the DET bracis-28398 141 53 purity purity NOUN bracis-28398 141 54 of of ADP bracis-28398 141 55 the the DET bracis-28398 141 56 footprint footprint NOUN bracis-28398 141 57 , , PUNCT bracis-28398 141 58 which which PRON bracis-28398 141 59 corresponds correspond VERB bracis-28398 141 60 to to ADP bracis-28398 141 61 the the DET bracis-28398 141 62 percentage percentage NOUN bracis-28398 141 63 of of ADP bracis-28398 141 64 instances instance NOUN bracis-28398 141 65 enclosed enclose VERB bracis-28398 141 66 by by ADP bracis-28398 141 67 the the DET bracis-28398 141 68 footprint footprint NOUN bracis-28398 141 69 from from ADP bracis-28398 141 70 a a DET bracis-28398 141 71 given give VERB bracis-28398 141 72 category category NOUN bracis-28398 141 73 , , PUNCT bracis-28398 141 74 that that ADV bracis-28398 141 75 is is ADV bracis-28398 141 76 , , PUNCT bracis-28398 141 77 the the DET bracis-28398 141 78 percentage percentage NOUN bracis-28398 141 79 of of ADP bracis-28398 141 80 hard hard ADJ bracis-28398 141 81 instances instance NOUN bracis-28398 141 82 in in ADP bracis-28398 141 83 the the DET bracis-28398 141 84 case case NOUN bracis-28398 141 85 of of ADP bracis-28398 141 86 the the DET bracis-28398 141 87 hardness hardness NOUN bracis-28398 141 88 footprint footprint NOUN bracis-28398 141 89 and and CCONJ bracis-28398 141 90 of of ADP bracis-28398 141 91 easy easy ADJ bracis-28398 141 92 instances instance NOUN bracis-28398 141 93 in in ADP bracis-28398 141 94 the the DET bracis-28398 141 95 easiness easiness ADJ bracis-28398 141 96 footprint footprint NOUN bracis-28398 141 97 . . PUNCT bracis-28398 142 1 taking take VERB bracis-28398 142 2 the the DET bracis-28398 142 3 hardness hardness ADJ bracis-28398 142 4 footprint footprint NOUN bracis-28398 142 5 as as ADP bracis-28398 142 6 reference reference NOUN bracis-28398 142 7 , , PUNCT bracis-28398 142 8 a a DET bracis-28398 142 9 large large ADJ bracis-28398 142 10 area area NOUN bracis-28398 142 11 implies imply VERB bracis-28398 142 12 the the DET bracis-28398 142 13 dataset dataset NOUN bracis-28398 142 14 has have VERB bracis-28398 142 15 many many ADJ bracis-28398 142 16 instances instance NOUN bracis-28398 142 17 hard hard ADJ bracis-28398 142 18 to to PART bracis-28398 142 19 classify classify VERB bracis-28398 142 20 , , PUNCT bracis-28398 142 21 so so SCONJ bracis-28398 142 22 that that SCONJ bracis-28398 142 23 they they PRON bracis-28398 142 24 occupy occupy VERB bracis-28398 142 25 a a DET bracis-28398 142 26 large large ADJ bracis-28398 142 27 portion portion NOUN bracis-28398 142 28 of of ADP bracis-28398 142 29 the the DET bracis-28398 142 30 hardness hardness NOUN bracis-28398 142 31 embedding embedding NOUN bracis-28398 142 32 . . PUNCT bracis-28398 143 1 a a DET bracis-28398 143 2 large large ADJ bracis-28398 143 3 density density NOUN bracis-28398 143 4 means mean VERB bracis-28398 143 5 such such ADJ bracis-28398 143 6 area area NOUN bracis-28398 143 7 is be AUX bracis-28398 143 8 dense dense ADJ bracis-28398 143 9 . . PUNCT bracis-28398 144 1 and and CCONJ bracis-28398 144 2 a a DET bracis-28398 144 3 large large ADJ bracis-28398 144 4 purity purity NOUN bracis-28398 144 5 is be AUX bracis-28398 144 6 observed observe VERB bracis-28398 144 7 when when SCONJ bracis-28398 144 8 most most ADJ bracis-28398 144 9 of of ADP bracis-28398 144 10 the the DET bracis-28398 144 11 instances instance NOUN bracis-28398 144 12 enclosed enclose VERB bracis-28398 144 13 in in ADP bracis-28398 144 14 the the DET bracis-28398 144 15 hardness hardness NOUN bracis-28398 144 16 footprint footprint NOUN bracis-28398 144 17 are be AUX bracis-28398 144 18 indeed indeed ADV bracis-28398 144 19 hard hard ADJ bracis-28398 144 20 . . PUNCT bracis-28398 145 1 conversely conversely ADV bracis-28398 145 2 , , PUNCT bracis-28398 145 3 it it PRON bracis-28398 145 4 is be AUX bracis-28398 145 5 possible possible ADJ bracis-28398 145 6 to to PART bracis-28398 145 7 define define VERB bracis-28398 145 8 similar similar ADJ bracis-28398 145 9 concepts concept NOUN bracis-28398 145 10 for for ADP bracis-28398 145 11 the the DET bracis-28398 145 12 easiness easiness ADJ bracis-28398 145 13 footprint footprint NOUN bracis-28398 145 14 and and CCONJ bracis-28398 145 15 the the DET bracis-28398 145 16 easy easy ADJ bracis-28398 145 17 instances instance NOUN bracis-28398 145 18 . . PUNCT bracis-28398 146 1 the the DET bracis-28398 146 2 easiness easiness NOUN bracis-28398 146 3 and and CCONJ bracis-28398 146 4 hardness hardness NOUN bracis-28398 146 5 footprints footprint NOUN bracis-28398 146 6 ’ ' PUNCT bracis-28398 146 7 areas area NOUN bracis-28398 146 8 can can AUX bracis-28398 146 9 be be AUX bracis-28398 146 10 regarded regard VERB bracis-28398 146 11 as as ADP bracis-28398 146 12 an an DET bracis-28398 146 13 indicative indicative NOUN bracis-28398 146 14 of of ADP bracis-28398 146 15 the the DET bracis-28398 146 16 hardness hardness NOUN bracis-28398 146 17 profile profile NOUN bracis-28398 146 18 of of ADP bracis-28398 146 19 the the DET bracis-28398 146 20 dataset dataset NOUN bracis-28398 146 21 . . PUNCT bracis-28398 147 1 when when SCONJ bracis-28398 147 2 the the DET bracis-28398 147 3 hardness hardness NOUN bracis-28398 147 4 area area NOUN bracis-28398 147 5 is be AUX bracis-28398 147 6 large large ADJ bracis-28398 147 7 , , PUNCT bracis-28398 147 8 the the DET bracis-28398 147 9 dataset dataset NOUN bracis-28398 147 10 and and CCONJ bracis-28398 147 11 the the DET bracis-28398 147 12 underlying underlie VERB bracis-28398 147 13 classification classification NOUN bracis-28398 147 14 problem problem NOUN bracis-28398 147 15 can can AUX bracis-28398 147 16 be be AUX bracis-28398 147 17 considered consider VERB bracis-28398 147 18 more more ADV bracis-28398 147 19 difficult difficult ADJ bracis-28398 147 20 to to PART bracis-28398 147 21 solve solve VERB bracis-28398 147 22 . . PUNCT bracis-28398 148 1 in in ADP bracis-28398 148 2 contrast contrast NOUN bracis-28398 148 3 , , PUNCT bracis-28398 148 4 easy easy ADJ bracis-28398 148 5 datasets dataset NOUN bracis-28398 148 6 / / SYM bracis-28398 148 7 problems problem NOUN bracis-28398 148 8 will will AUX bracis-28398 148 9 show show VERB bracis-28398 148 10 a a DET bracis-28398 148 11 large large ADJ bracis-28398 148 12 easiness easiness ADJ bracis-28398 148 13 area area NOUN bracis-28398 148 14 . . PUNCT bracis-28398 149 1 4 4 NUM bracis-28398 149 2 experiments experiment NOUN bracis-28398 149 3 four four NUM bracis-28398 149 4 datasets dataset NOUN bracis-28398 149 5 from from ADP bracis-28398 149 6 the the DET bracis-28398 149 7 health health NOUN bracis-28398 149 8 domain domain NOUN bracis-28398 149 9 , , PUNCT bracis-28398 149 10 involving involve VERB bracis-28398 149 11 the the DET bracis-28398 149 12 prognosis prognosis NOUN bracis-28398 149 13 of of ADP bracis-28398 149 14 individuals individual NOUN bracis-28398 149 15 diagnosed diagnose VERB bracis-28398 149 16 for for ADP bracis-28398 149 17 covid-19 covid-19 PROPN bracis-28398 149 18 in in ADP bracis-28398 149 19 brazil brazil PROPN bracis-28398 149 20 , , PUNCT bracis-28398 149 21 were be AUX bracis-28398 149 22 used use VERB bracis-28398 149 23 in in ADP bracis-28398 149 24 our our PRON bracis-28398 149 25 experiments experiment NOUN bracis-28398 149 26 . . PUNCT bracis-28398 150 1 datasets dataset NOUN bracis-28398 150 2 and and CCONJ bracis-28398 150 3 their their PRON bracis-28398 150 4 description description NOUN bracis-28398 150 5 are be AUX bracis-28398 150 6 available available ADJ bracis-28398 150 7 in in ADP bracis-28398 150 8 a a DET bracis-28398 150 9 public public ADJ bracis-28398 150 10 repositoryfootnote repositoryfootnote NOUN bracis-28398 150 11 1 1 NUM bracis-28398 150 12 . . X bracis-28398 151 1 summarizing summarizing NOUN bracis-28398 151 2 , , PUNCT bracis-28398 151 3 we we PRON bracis-28398 151 4 have have VERB bracis-28398 151 5 : : PUNCT bracis-28398 151 6 1 1 X bracis-28398 151 7 . . X bracis-28398 151 8 hospital hospital NOUN bracis-28398 151 9 1 1 NUM bracis-28398 151 10 and and CCONJ bracis-28398 151 11 hospital hospital NOUN bracis-28398 151 12 2 2 NUM bracis-28398 151 13 : : PUNCT bracis-28398 151 14 prognosis prognosis NOUN bracis-28398 151 15 of of ADP bracis-28398 151 16 covid-19 covid-19 PROPN bracis-28398 151 17 severity severity NOUN bracis-28398 151 18 for for ADP bracis-28398 151 19 patients patient NOUN bracis-28398 151 20 hospitalized hospitalize VERB bracis-28398 151 21 at at ADP bracis-28398 151 22 two two NUM bracis-28398 151 23 distinct distinct ADJ bracis-28398 151 24 private private ADJ bracis-28398 151 25 hospitals hospital NOUN bracis-28398 151 26 from from ADP bracis-28398 151 27 the the DET bracis-28398 151 28 são são PROPN bracis-28398 151 29 paulo paulo PROPN bracis-28398 151 30 metropolitan metropolitan PROPN bracis-28398 151 31 area area PROPN bracis-28398 151 32 , , PUNCT bracis-28398 151 33 using use VERB bracis-28398 151 34 some some DET bracis-28398 151 35 standard standard ADJ bracis-28398 151 36 laboratory laboratory NOUN bracis-28398 151 37 tests test NOUN bracis-28398 151 38 ; ; PUNCT bracis-28398 151 39 2 2 X bracis-28398 151 40 . . X bracis-28398 151 41 hospitalization hospitalization NOUN bracis-28398 151 42 and and CCONJ bracis-28398 151 43 severity severity NOUN bracis-28398 151 44 : : PUNCT bracis-28398 151 45 prediction prediction NOUN bracis-28398 151 46 of of ADP bracis-28398 151 47 the the DET bracis-28398 151 48 hospitalization hospitalization NOUN bracis-28398 151 49 length length NOUN bracis-28398 151 50 when when SCONJ bracis-28398 151 51 a a DET bracis-28398 151 52 citizen citizen NOUN bracis-28398 151 53 from from ADP bracis-28398 151 54 são são PROPN bracis-28398 151 55 josé josé PROPN bracis-28398 151 56 dos dos PROPN bracis-28398 151 57 campos campos PROPN bracis-28398 151 58 - - PUNCT bracis-28398 151 59 sp sp NOUN bracis-28398 151 60 is be AUX bracis-28398 151 61 hospitalized hospitalize VERB bracis-28398 151 62 with with ADP bracis-28398 151 63 a a DET bracis-28398 151 64 positive positive ADJ bracis-28398 151 65 covid covid NOUN bracis-28398 151 66 diagnosis diagnosis NOUN bracis-28398 151 67 ( ( PUNCT bracis-28398 151 68 hospitalization hospitalization NOUN bracis-28398 151 69 ) ) PUNCT bracis-28398 151 70 and and CCONJ bracis-28398 151 71 if if SCONJ bracis-28398 151 72 he he PRON bracis-28398 151 73 / / PUNCT bracis-28398 151 74 she she PRON bracis-28398 151 75 will will AUX bracis-28398 151 76 evolve evolve VERB bracis-28398 151 77 to to ADP bracis-28398 151 78 a a DET bracis-28398 151 79 severe severe ADJ bracis-28398 151 80 condition condition NOUN bracis-28398 151 81 ( ( PUNCT bracis-28398 151 82 severity severity NOUN bracis-28398 151 83 ) ) PUNCT bracis-28398 151 84 , , PUNCT bracis-28398 151 85 based base VERB bracis-28398 151 86 on on ADP bracis-28398 151 87 initial initial ADJ bracis-28398 151 88 symptoms symptom NOUN bracis-28398 151 89 and and CCONJ bracis-28398 151 90 comorbidities comorbiditie NOUN bracis-28398 151 91 reported report VERB bracis-28398 151 92 . . PUNCT bracis-28398 152 1 4.1 4.1 NUM bracis-28398 152 2 footprint footprint NOUN bracis-28398 152 3 analysis analysis NOUN bracis-28398 152 4 first first ADV bracis-28398 152 5 , , PUNCT bracis-28398 152 6 we we PRON bracis-28398 152 7 analyze analyze VERB bracis-28398 152 8 the the DET bracis-28398 152 9 easiness easiness NOUN bracis-28398 152 10 and and CCONJ bracis-28398 152 11 hardness hardness ADJ bracis-28398 152 12 footprints footprint NOUN bracis-28398 152 13 of of ADP bracis-28398 152 14 three three NUM bracis-28398 152 15 datasets dataset NOUN bracis-28398 152 16 with with ADP bracis-28398 152 17 distinct distinct ADJ bracis-28398 152 18 hardness hardness ADJ bracis-28398 152 19 profiles profile NOUN bracis-28398 152 20 . . PUNCT bracis-28398 153 1 they they PRON bracis-28398 153 2 are be AUX bracis-28398 153 3 severity severity NOUN bracis-28398 153 4 , , PUNCT bracis-28398 153 5 hospital hospital NOUN bracis-28398 153 6 1 1 NUM bracis-28398 153 7 , , PUNCT bracis-28398 153 8 and and CCONJ bracis-28398 153 9 hospitalization hospitalization NOUN bracis-28398 153 10 datasets dataset NOUN bracis-28398 153 11 with with ADP bracis-28398 153 12 easy easy ADJ bracis-28398 153 13 , , PUNCT bracis-28398 153 14 intermediate intermediate ADJ bracis-28398 153 15 , , PUNCT bracis-28398 153 16 and and CCONJ bracis-28398 153 17 hard hard ADJ bracis-28398 153 18 profiles profile NOUN bracis-28398 153 19 , , PUNCT bracis-28398 153 20 respectively respectively ADV bracis-28398 153 21 . . PUNCT bracis-28398 154 1 following follow VERB bracis-28398 154 2 the the DET bracis-28398 154 3 same same ADJ bracis-28398 154 4 order order NOUN bracis-28398 154 5 , , PUNCT bracis-28398 154 6 the the DET bracis-28398 154 7 average average ADJ bracis-28398 154 8 accuracy accuracy NOUN bracis-28398 154 9 rate rate NOUN bracis-28398 154 10 achieved achieve VERB bracis-28398 154 11 by by ADP bracis-28398 154 12 the the DET bracis-28398 154 13 seven seven NUM bracis-28398 154 14 classifiers classifier NOUN bracis-28398 154 15 for for ADP bracis-28398 154 16 each each PRON bracis-28398 154 17 of of ADP bracis-28398 154 18 these these DET bracis-28398 154 19 datasets dataset NOUN bracis-28398 154 20 was be AUX bracis-28398 154 21 : : PUNCT bracis-28398 154 22 0.900 0.900 NUM bracis-28398 154 23 ( ( PUNCT bracis-28398 154 24 std std NOUN bracis-28398 154 25 0.003 0.003 NUM bracis-28398 154 26 ) ) PUNCT bracis-28398 154 27 , , PUNCT bracis-28398 154 28 0.673 0.673 NUM bracis-28398 154 29 ( ( PUNCT bracis-28398 154 30 std std NOUN bracis-28398 154 31 0.030 0.030 NUM bracis-28398 154 32 ) ) PUNCT bracis-28398 154 33 , , PUNCT bracis-28398 154 34 and and CCONJ bracis-28398 154 35 0.398 0.398 NUM bracis-28398 154 36 ( ( PUNCT bracis-28398 154 37 std std NOUN bracis-28398 154 38 0.022 0.022 NUM bracis-28398 154 39 ) ) PUNCT bracis-28398 154 40 . . PUNCT bracis-28398 155 1 fig fig NOUN bracis-28398 155 2 . . PUNCT bracis-28398 156 1 3 3 X bracis-28398 156 2 . . NOUN bracis-28398 156 3 easy easy ADJ bracis-28398 156 4 and and CCONJ bracis-28398 156 5 hard hard ADJ bracis-28398 156 6 footprints footprint NOUN bracis-28398 156 7 of of ADP bracis-28398 156 8 datasets dataset NOUN bracis-28398 156 9 with with ADP bracis-28398 156 10 contrasting contrast VERB bracis-28398 156 11 profiles profile NOUN bracis-28398 156 12 : : PUNCT bracis-28398 156 13 ( ( PUNCT bracis-28398 156 14 a a X bracis-28398 156 15 ) ) PUNCT bracis-28398 156 16 severity severity NOUN bracis-28398 156 17 , , PUNCT bracis-28398 156 18 ( ( PUNCT bracis-28398 156 19 b b X bracis-28398 156 20 ) ) PUNCT bracis-28398 156 21 hospital hospital NOUN bracis-28398 156 22 1 1 NUM bracis-28398 156 23 ; ; PUNCT bracis-28398 156 24 ( ( PUNCT bracis-28398 156 25 c c X bracis-28398 156 26 ) ) PUNCT bracis-28398 156 27 hospitalization hospitalization NOUN bracis-28398 156 28 . . PUNCT bracis-28398 157 1 t t NOUN bracis-28398 157 2 value value NOUN bracis-28398 157 3 = = SYM bracis-28398 157 4 0.048 0.048 NUM bracis-28398 157 5 . . PUNCT bracis-28398 158 1 full full ADJ bracis-28398 158 2 size size NOUN bracis-28398 158 3 image image NOUN bracis-28398 158 4 to to PART bracis-28398 158 5 determine determine VERB bracis-28398 158 6 such such ADJ bracis-28398 158 7 footprints footprint NOUN bracis-28398 158 8 , , PUNCT bracis-28398 158 9 we we PRON bracis-28398 158 10 did do VERB bracis-28398 158 11 an an DET bracis-28398 158 12 experimental experimental ADJ bracis-28398 158 13 procedure procedure NOUN bracis-28398 158 14 varying vary VERB bracis-28398 158 15 the the DET bracis-28398 158 16 t t NOUN bracis-28398 158 17 value value NOUN bracis-28398 158 18 from from ADP bracis-28398 158 19 0.3 0.3 NUM bracis-28398 158 20 to to ADP bracis-28398 158 21 0.495 0.495 NUM bracis-28398 158 22 , , PUNCT bracis-28398 158 23 with with ADP bracis-28398 158 24 steps step NOUN bracis-28398 158 25 of of ADP bracis-28398 158 26 0.02 0.02 NUM bracis-28398 158 27 . . PUNCT bracis-28398 159 1 next next ADJ bracis-28398 159 2 , , PUNCT bracis-28398 159 3 we we PRON bracis-28398 159 4 monitored monitor VERB bracis-28398 159 5 the the DET bracis-28398 159 6 purity purity NOUN bracis-28398 159 7 values value NOUN bracis-28398 159 8 of of ADP bracis-28398 159 9 the the DET bracis-28398 159 10 resulting result VERB bracis-28398 159 11 footprints footprint NOUN bracis-28398 159 12 . . PUNCT bracis-28398 160 1 we we PRON bracis-28398 160 2 chose choose VERB bracis-28398 160 3 the the DET bracis-28398 160 4 t t PROPN bracis-28398 160 5 values value NOUN bracis-28398 160 6 leading lead VERB bracis-28398 160 7 to to ADP bracis-28398 160 8 the the DET bracis-28398 160 9 highest high ADJ bracis-28398 160 10 purity purity NOUN bracis-28398 160 11 values value NOUN bracis-28398 160 12 for for ADP bracis-28398 160 13 both both CCONJ bracis-28398 160 14 easiness easiness NOUN bracis-28398 160 15 and and CCONJ bracis-28398 160 16 hardness hardness NOUN bracis-28398 160 17 footprints footprint NOUN bracis-28398 160 18 . . PUNCT bracis-28398 161 1 purest pure ADJ bracis-28398 161 2 footprints footprint NOUN bracis-28398 161 3 are be AUX bracis-28398 161 4 preferred prefer VERB bracis-28398 161 5 , , PUNCT bracis-28398 161 6 as as SCONJ bracis-28398 161 7 they they PRON bracis-28398 161 8 ensure ensure VERB bracis-28398 161 9 most most ADJ bracis-28398 161 10 of of ADP bracis-28398 161 11 the the DET bracis-28398 161 12 instances instance NOUN bracis-28398 161 13 the the DET bracis-28398 161 14 footprint footprint NOUN bracis-28398 161 15 encloses enclose NOUN bracis-28398 161 16 are be AUX bracis-28398 161 17 indeed indeed ADV bracis-28398 161 18 easy easy ADV bracis-28398 161 19 / / SYM bracis-28398 161 20 hard hard ADJ bracis-28398 161 21 . . PUNCT bracis-28398 162 1 figure figure NOUN bracis-28398 162 2     SPACE bracis-28398 162 3 3 3 NUM bracis-28398 162 4 shows show VERB bracis-28398 162 5 the the DET bracis-28398 162 6 obtained obtain VERB bracis-28398 162 7 footprints footprint NOUN bracis-28398 162 8 : : PUNCT bracis-28398 162 9 hardness hardness ADJ bracis-28398 162 10 in in ADP bracis-28398 162 11 red red ADJ bracis-28398 162 12 and and CCONJ bracis-28398 162 13 easiness easiness NOUN bracis-28398 162 14 in in ADP bracis-28398 162 15 blue blue ADJ bracis-28398 162 16 . . PUNCT bracis-28398 163 1 as as SCONJ bracis-28398 163 2 expected expect VERB bracis-28398 163 3 , , PUNCT bracis-28398 163 4 the the DET bracis-28398 163 5 hardness hardness ADJ bracis-28398 163 6 footprint footprint NOUN bracis-28398 163 7 dimensions dimension NOUN bracis-28398 163 8 increase increase NOUN bracis-28398 163 9 according accord VERB bracis-28398 163 10 to to ADP bracis-28398 163 11 the the DET bracis-28398 163 12 dataset dataset NOUN bracis-28398 163 13 ’s ’s PART bracis-28398 163 14 difficulty difficulty NOUN bracis-28398 163 15 level level NOUN bracis-28398 163 16 . . PUNCT bracis-28398 164 1 for for ADP bracis-28398 164 2 instance instance NOUN bracis-28398 164 3 , , PUNCT bracis-28398 164 4 when when SCONJ bracis-28398 164 5 the the DET bracis-28398 164 6 t t NOUN bracis-28398 164 7 value value NOUN bracis-28398 164 8 reaches reach VERB bracis-28398 164 9 0.45 0.45 NUM bracis-28398 164 10 , , PUNCT bracis-28398 164 11 all all DET bracis-28398 164 12 instances instance NOUN bracis-28398 164 13 are be AUX bracis-28398 164 14 considered consider VERB bracis-28398 164 15 hard hard ADJ bracis-28398 164 16 in in ADP bracis-28398 164 17 the the DET bracis-28398 164 18 hospitalization hospitalization NOUN bracis-28398 164 19 dataset dataset NOUN bracis-28398 164 20 . . PUNCT bracis-28398 165 1 fig fig NOUN bracis-28398 165 2 . . PUNCT bracis-28398 166 1 4 4 X bracis-28398 166 2 . . X bracis-28398 166 3 areas area NOUN bracis-28398 166 4 , , PUNCT bracis-28398 166 5 purities purity NOUN bracis-28398 166 6 and and CCONJ bracis-28398 166 7 densities density NOUN bracis-28398 166 8 of of ADP bracis-28398 166 9 the the DET bracis-28398 166 10 hardness hardness ADJ bracis-28398 166 11 footprints footprint NOUN bracis-28398 166 12 for for ADP bracis-28398 166 13 increasing increase VERB bracis-28398 166 14 t t PROPN bracis-28398 166 15 values value NOUN bracis-28398 166 16 . . PUNCT bracis-28398 167 1 full full ADJ bracis-28398 167 2 size size NOUN bracis-28398 167 3 image image NOUN bracis-28398 167 4 figure figure NOUN bracis-28398 167 5     SPACE bracis-28398 167 6 4 4 NUM bracis-28398 167 7 presents present VERB bracis-28398 167 8 the the DET bracis-28398 167 9 variations variation NOUN bracis-28398 167 10 of of ADP bracis-28398 167 11 the the DET bracis-28398 167 12 area area NOUN bracis-28398 167 13 , , PUNCT bracis-28398 167 14 density density NOUN bracis-28398 167 15 and and CCONJ bracis-28398 167 16 purity purity NOUN bracis-28398 167 17 values value NOUN bracis-28398 167 18 of of ADP bracis-28398 167 19 the the DET bracis-28398 167 20 hardness hardness NOUN bracis-28398 167 21 footprint footprint NOUN bracis-28398 167 22 for for ADP bracis-28398 167 23 increasing increase VERB bracis-28398 167 24 t t PROPN bracis-28398 167 25 values value NOUN bracis-28398 167 26 in in ADP bracis-28398 167 27 the the DET bracis-28398 167 28 datasets dataset NOUN bracis-28398 167 29 . . PUNCT bracis-28398 168 1 as as SCONJ bracis-28398 168 2 expected expect VERB bracis-28398 168 3 , , PUNCT bracis-28398 168 4 when when SCONJ bracis-28398 168 5 t t PROPN bracis-28398 168 6 increases increase VERB bracis-28398 168 7 , , PUNCT bracis-28398 168 8 the the DET bracis-28398 168 9 footprints footprint NOUN bracis-28398 168 10 ’ ' PUNCT bracis-28398 168 11 areas area NOUN bracis-28398 168 12 grow grow VERB bracis-28398 168 13 , , PUNCT bracis-28398 168 14 as as SCONJ bracis-28398 168 15 more more ADJ bracis-28398 168 16 instances instance NOUN bracis-28398 168 17 obey obey VERB bracis-28398 168 18 the the DET bracis-28398 168 19 inequalities inequality NOUN bracis-28398 168 20 involving involve VERB bracis-28398 168 21 the the DET bracis-28398 168 22 instance instance NOUN bracis-28398 168 23 hardness hardness NOUN bracis-28398 168 24 values value NOUN bracis-28398 168 25 . . PUNCT bracis-28398 169 1 but but CCONJ bracis-28398 169 2 this this DET bracis-28398 169 3 increase increase NOUN bracis-28398 169 4 is be AUX bracis-28398 169 5 less less ADV bracis-28398 169 6 accentuated accentuated ADJ bracis-28398 169 7 for for ADP bracis-28398 169 8 the the DET bracis-28398 169 9 datasets dataset NOUN bracis-28398 169 10 with with ADP bracis-28398 169 11 easy easy ADJ bracis-28398 169 12 and and CCONJ bracis-28398 169 13 intermediate intermediate ADJ bracis-28398 169 14 hardness hardness ADJ bracis-28398 169 15 profiles profile NOUN bracis-28398 169 16 . . PUNCT bracis-28398 170 1 purity purity NOUN bracis-28398 170 2 is be AUX bracis-28398 170 3 always always ADV bracis-28398 170 4 over over ADP bracis-28398 170 5 0.7 0.7 NUM bracis-28398 170 6 and and CCONJ bracis-28398 170 7 varies vary VERB bracis-28398 170 8 less less ADJ bracis-28398 170 9 for for ADP bracis-28398 170 10 the the DET bracis-28398 170 11 severity severity NOUN bracis-28398 170 12 dataset dataset NOUN bracis-28398 170 13 , , PUNCT bracis-28398 170 14 while while SCONJ bracis-28398 170 15 presenting present VERB bracis-28398 170 16 an an DET bracis-28398 170 17 increasing increase VERB bracis-28398 170 18 tendency tendency NOUN bracis-28398 170 19 in in ADP bracis-28398 170 20 the the DET bracis-28398 170 21 other other ADJ bracis-28398 170 22 problems problem NOUN bracis-28398 170 23 . . PUNCT bracis-28398 171 1 in in ADP bracis-28398 171 2 contrast contrast NOUN bracis-28398 171 3 , , PUNCT bracis-28398 171 4 whilst whilst SCONJ bracis-28398 171 5 density density NOUN bracis-28398 171 6 values value NOUN bracis-28398 171 7 are be AUX bracis-28398 171 8 stable stable ADJ bracis-28398 171 9 in in ADP bracis-28398 171 10 the the DET bracis-28398 171 11 hospitalization hospitalization NOUN bracis-28398 171 12 dataset dataset NOUN bracis-28398 171 13 , , PUNCT bracis-28398 171 14 where where SCONJ bracis-28398 171 15 the the DET bracis-28398 171 16 instances instance NOUN bracis-28398 171 17 are be AUX bracis-28398 171 18 concentrated concentrate VERB bracis-28398 171 19 in in ADP bracis-28398 171 20 the the DET bracis-28398 171 21 center center NOUN bracis-28398 171 22 of of ADP bracis-28398 171 23 the the PRON bracis-28398 171 24 is be AUX bracis-28398 171 25 , , PUNCT bracis-28398 171 26 a a DET bracis-28398 171 27 tendency tendency NOUN bracis-28398 171 28 of of ADP bracis-28398 171 29 decrease decrease NOUN bracis-28398 171 30 is be AUX bracis-28398 171 31 observed observe VERB bracis-28398 171 32 for for ADP bracis-28398 171 33 the the DET bracis-28398 171 34 other other ADJ bracis-28398 171 35 datasets dataset NOUN bracis-28398 171 36 ( ( PUNCT bracis-28398 171 37 larger large ADJ bracis-28398 171 38 areas area NOUN bracis-28398 171 39 with with ADP bracis-28398 171 40 less less ADJ bracis-28398 171 41 instances instance NOUN bracis-28398 171 42 are be AUX bracis-28398 171 43 encompassed encompass VERB bracis-28398 171 44 in in ADP bracis-28398 171 45 the the DET bracis-28398 171 46 footprints footprint NOUN bracis-28398 171 47 ) ) PUNCT bracis-28398 171 48 . . PUNCT bracis-28398 172 1 similar similar ADJ bracis-28398 172 2 plots plot NOUN bracis-28398 172 3 can can AUX bracis-28398 172 4 be be AUX bracis-28398 172 5 generated generate VERB bracis-28398 172 6 for for ADP bracis-28398 172 7 the the DET bracis-28398 172 8 easiness easiness ADJ bracis-28398 172 9 footprints footprint NOUN bracis-28398 172 10 . . PUNCT bracis-28398 173 1 from from ADP bracis-28398 173 2 the the DET bracis-28398 173 3 footprints footprint NOUN bracis-28398 173 4 we we PRON bracis-28398 173 5 can can AUX bracis-28398 173 6 already already ADV bracis-28398 173 7 obtain obtain VERB bracis-28398 173 8 insights insight NOUN bracis-28398 173 9 by by ADP bracis-28398 173 10 contrasting contrast VERB bracis-28398 173 11 hard hard ADJ bracis-28398 173 12 and and CCONJ bracis-28398 173 13 easy easy ADJ bracis-28398 173 14 instances instance NOUN bracis-28398 173 15 . . PUNCT bracis-28398 174 1 in in ADP bracis-28398 174 2 fig fig NOUN bracis-28398 174 3 . . PUNCT bracis-28398 174 4     SPACE bracis-28398 174 5 5 5 NUM bracis-28398 174 6 we we PRON bracis-28398 174 7 present present VERB bracis-28398 174 8 boxplots boxplot NOUN bracis-28398 174 9 contrasting contrast VERB bracis-28398 174 10 the the DET bracis-28398 174 11 percentage percentage NOUN bracis-28398 174 12 of of ADP bracis-28398 174 13 lymphocytes lymphocyte NOUN bracis-28398 174 14 , , PUNCT bracis-28398 174 15 one one NUM bracis-28398 174 16 input input NOUN bracis-28398 174 17 feature feature NOUN bracis-28398 174 18 of of ADP bracis-28398 174 19 dataset dataset ADJ bracis-28398 174 20 hospital hospital NOUN bracis-28398 174 21 1 1 NUM bracis-28398 174 22 for for ADP bracis-28398 174 23 instances instance NOUN bracis-28398 174 24 in in ADP bracis-28398 174 25 the the DET bracis-28398 174 26 hardness hardness NOUN bracis-28398 174 27 ( ( PUNCT bracis-28398 174 28 left leave VERB bracis-28398 174 29 boxplots boxplot NOUN bracis-28398 174 30 ) ) PUNCT bracis-28398 174 31 and and CCONJ bracis-28398 174 32 easiness easiness NOUN bracis-28398 174 33 ( ( PUNCT bracis-28398 174 34 right right ADJ bracis-28398 174 35 boxplots boxplot NOUN bracis-28398 174 36 ) ) PUNCT bracis-28398 174 37 footprints footprint NOUN bracis-28398 174 38 . . PUNCT bracis-28398 175 1 colors color NOUN bracis-28398 175 2 matches match VERB bracis-28398 175 3 the the DET bracis-28398 175 4 original original ADJ bracis-28398 175 5 class class NOUN bracis-28398 175 6 , , PUNCT bracis-28398 175 7 where where SCONJ bracis-28398 175 8 red red PROPN bracis-28398 175 9 stands stand VERB bracis-28398 175 10 for for ADP bracis-28398 175 11 patients patient NOUN bracis-28398 175 12 that that PRON bracis-28398 175 13 developed develop VERB bracis-28398 175 14 to to ADP bracis-28398 175 15 a a DET bracis-28398 175 16 severe severe ADJ bracis-28398 175 17 condition condition NOUN bracis-28398 175 18 and and CCONJ bracis-28398 175 19 blue blue NOUN bracis-28398 175 20 is be AUX bracis-28398 175 21 the the DET bracis-28398 175 22 opposite opposite ADJ bracis-28398 175 23 class class NOUN bracis-28398 175 24 . . PUNCT bracis-28398 176 1 clearly clearly ADV bracis-28398 176 2 , , PUNCT bracis-28398 176 3 there there PRON bracis-28398 176 4 is be VERB bracis-28398 176 5 an an DET bracis-28398 176 6 inversion inversion NOUN bracis-28398 176 7 on on ADP bracis-28398 176 8 the the DET bracis-28398 176 9 expected expect VERB bracis-28398 176 10 features feature NOUN bracis-28398 176 11 ’ ' PUNCT bracis-28398 176 12 values value NOUN bracis-28398 176 13 when when SCONJ bracis-28398 176 14 we we PRON bracis-28398 176 15 compare compare VERB bracis-28398 176 16 the the DET bracis-28398 176 17 plots plot NOUN bracis-28398 176 18 . . PUNCT bracis-28398 177 1 in in ADP bracis-28398 177 2 fact fact NOUN bracis-28398 177 3 , , PUNCT bracis-28398 177 4 a a DET bracis-28398 177 5 reduction reduction NOUN bracis-28398 177 6 in in ADP bracis-28398 177 7 the the DET bracis-28398 177 8 proportion proportion NOUN bracis-28398 177 9 of of ADP bracis-28398 177 10 lymphocytes lymphocyte NOUN bracis-28398 177 11 is be AUX bracis-28398 177 12 expected expect VERB bracis-28398 177 13 for for ADP bracis-28398 177 14 severe severe ADJ bracis-28398 177 15 cases case NOUN bracis-28398 177 16 and and CCONJ bracis-28398 177 17 not not PART bracis-28398 177 18 the the DET bracis-28398 177 19 opposite opposite ADJ bracis-28398 177 20     SPACE bracis-28398 177 21 [ [ X bracis-28398 177 22 20 20 NUM bracis-28398 177 23 ] ] PUNCT bracis-28398 177 24 . . PUNCT bracis-28398 178 1 this this PRON bracis-28398 178 2 is be AUX bracis-28398 178 3 already already ADV bracis-28398 178 4 an an DET bracis-28398 178 5 indicative indicative ADJ bracis-28398 178 6 that that SCONJ bracis-28398 178 7 hard hard ADJ bracis-28398 178 8 instances instance NOUN bracis-28398 178 9 are be AUX bracis-28398 178 10 not not PART bracis-28398 178 11 following follow VERB bracis-28398 178 12 the the DET bracis-28398 178 13 expected expect VERB bracis-28398 178 14 pattern pattern NOUN bracis-28398 178 15 from from ADP bracis-28398 178 16 the the DET bracis-28398 178 17 class class NOUN bracis-28398 178 18 they they PRON bracis-28398 178 19 belong belong VERB bracis-28398 178 20 to to ADP bracis-28398 178 21 . . PUNCT bracis-28398 179 1 figure figure VERB bracis-28398 179 2     SPACE bracis-28398 179 3 5b 5b PROPN bracis-28398 179 4 presents present NOUN bracis-28398 179 5 boxplots boxplot NOUN bracis-28398 179 6 of of ADP bracis-28398 179 7 the the DET bracis-28398 179 8 kdn kdn NOUN bracis-28398 179 9 meta meta ADJ bracis-28398 179 10 - - PUNCT bracis-28398 179 11 feature feature NOUN bracis-28398 179 12 values value NOUN bracis-28398 179 13 for for ADP bracis-28398 179 14 the the DET bracis-28398 179 15 hardness hardness NOUN bracis-28398 179 16 ( ( PUNCT bracis-28398 179 17 left left ADJ bracis-28398 179 18 ) ) PUNCT bracis-28398 179 19 and and CCONJ bracis-28398 179 20 easiness easiness NOUN bracis-28398 179 21 ( ( PUNCT bracis-28398 179 22 right right ADJ bracis-28398 179 23 ) ) PUNCT bracis-28398 179 24 footprints footprint NOUN bracis-28398 179 25 . . PUNCT bracis-28398 180 1 easy easy ADJ bracis-28398 180 2 instances instance NOUN bracis-28398 180 3 have have VERB bracis-28398 180 4 a a DET bracis-28398 180 5 low low ADJ bracis-28398 180 6 kdn kdn NOUN bracis-28398 180 7 value value NOUN bracis-28398 180 8 ( ( PUNCT bracis-28398 180 9 are be AUX bracis-28398 180 10 surrounded surround VERB bracis-28398 180 11 by by ADP bracis-28398 180 12 elements element NOUN bracis-28398 180 13 sharing share VERB bracis-28398 180 14 their their PRON bracis-28398 180 15 class class NOUN bracis-28398 180 16 label label NOUN bracis-28398 180 17 ) ) PUNCT bracis-28398 180 18 , , PUNCT bracis-28398 180 19 while while SCONJ bracis-28398 180 20 hard hard ADJ bracis-28398 180 21 instances instance NOUN bracis-28398 180 22 have have VERB bracis-28398 180 23 a a DET bracis-28398 180 24 high high ADJ bracis-28398 180 25 kdn kdn NOUN bracis-28398 180 26 value value NOUN bracis-28398 180 27 ( ( PUNCT bracis-28398 180 28 are be AUX bracis-28398 180 29 close close ADJ bracis-28398 180 30 to to ADP bracis-28398 180 31 elements element NOUN bracis-28398 180 32 from from ADP bracis-28398 180 33 the the DET bracis-28398 180 34 opposite opposite ADJ bracis-28398 180 35 class class NOUN bracis-28398 180 36 ) ) PUNCT bracis-28398 180 37 , , PUNCT bracis-28398 180 38 despite despite SCONJ bracis-28398 180 39 the the DET bracis-28398 180 40 class class NOUN bracis-28398 180 41 they they PRON bracis-28398 180 42 belong belong VERB bracis-28398 180 43 to to ADP bracis-28398 180 44 . . PUNCT bracis-28398 181 1 therefore therefore ADV bracis-28398 181 2 , , PUNCT bracis-28398 181 3 hard hard ADJ bracis-28398 181 4 instances instance NOUN bracis-28398 181 5 are be AUX bracis-28398 181 6 contrasting contrast VERB bracis-28398 181 7 their their PRON bracis-28398 181 8 ground ground NOUN bracis-28398 181 9 truth truth NOUN bracis-28398 181 10 labels label NOUN bracis-28398 181 11 and and CCONJ bracis-28398 181 12 are be AUX bracis-28398 181 13 contained contain VERB bracis-28398 181 14 in in ADP bracis-28398 181 15 overlapping overlap VERB bracis-28398 181 16 areas area NOUN bracis-28398 181 17 . . PUNCT bracis-28398 182 1 fig fig NOUN bracis-28398 182 2 . . PUNCT bracis-28398 183 1 5 5 X bracis-28398 183 2 . . X bracis-28398 183 3 plots plot NOUN bracis-28398 183 4 of of ADP bracis-28398 183 5 the the DET bracis-28398 183 6 input input NOUN bracis-28398 183 7 features feature VERB bracis-28398 183 8 lymphocytes lymphocyte VERB bracis-28398 183 9 ( ( PUNCT bracis-28398 183 10 % % INTJ bracis-28398 183 11 ) ) PUNCT bracis-28398 183 12 and and CCONJ bracis-28398 183 13 of of ADP bracis-28398 183 14 meta meta ADJ bracis-28398 183 15 - - PUNCT bracis-28398 183 16 feature feature NOUN bracis-28398 183 17 kdn kdn NOUN bracis-28398 183 18 for for ADP bracis-28398 183 19 the the DET bracis-28398 183 20 hard hard ADJ bracis-28398 183 21 instances instance NOUN bracis-28398 183 22 ( ( PUNCT bracis-28398 183 23 left leave VERB bracis-28398 183 24 ) ) PUNCT bracis-28398 183 25 and and CCONJ bracis-28398 183 26 easy easy ADJ bracis-28398 183 27 instances instance NOUN bracis-28398 183 28 ( ( PUNCT bracis-28398 183 29 right right ADJ bracis-28398 183 30 ) ) PUNCT bracis-28398 183 31 from from ADP bracis-28398 183 32 dataset dataset ADJ bracis-28398 183 33 hospital hospital NOUN bracis-28398 183 34 1 1 NUM bracis-28398 183 35 . . PUNCT bracis-28398 183 36 full full ADJ bracis-28398 183 37 size size NOUN bracis-28398 183 38 image image NOUN bracis-28398 183 39 4.2 4.2 NUM bracis-28398 183 40 data datum NOUN bracis-28398 183 41 sculpting sculpt VERB bracis-28398 183 42 fig fig NOUN bracis-28398 183 43 . . PUNCT bracis-28398 184 1 6 6 NUM bracis-28398 184 2 . . X bracis-28398 184 3 average average ADJ bracis-28398 184 4 model model NOUN bracis-28398 184 5 performance performance NOUN bracis-28398 184 6 scored score VERB bracis-28398 184 7 by by ADP bracis-28398 184 8 auc auc NOUN bracis-28398 184 9 in in ADP bracis-28398 184 10 a a DET bracis-28398 184 11 data data NOUN bracis-28398 184 12 sculpting sculpting NOUN bracis-28398 184 13 process process NOUN bracis-28398 184 14 . . PUNCT bracis-28398 185 1 the the DET bracis-28398 185 2 dotted dotted ADJ bracis-28398 185 3 line line NOUN bracis-28398 185 4 is be AUX bracis-28398 185 5 the the DET bracis-28398 185 6 performance performance NOUN bracis-28398 185 7 without without ADP bracis-28398 185 8 removing remove VERB bracis-28398 185 9 any any DET bracis-28398 185 10 instance instance NOUN bracis-28398 185 11 from from ADP bracis-28398 185 12 hospital hospital NOUN bracis-28398 185 13 1 1 NUM bracis-28398 185 14 . . PUNCT bracis-28398 185 15 full full ADJ bracis-28398 185 16 size size NOUN bracis-28398 185 17 image image NOUN bracis-28398 185 18 now now ADV bracis-28398 185 19 we we PRON bracis-28398 185 20 move move VERB bracis-28398 185 21 our our PRON bracis-28398 185 22 analysis analysis NOUN bracis-28398 185 23 to to ADP bracis-28398 185 24 a a DET bracis-28398 185 25 practical practical ADJ bracis-28398 185 26 usage usage NOUN bracis-28398 185 27 of of ADP bracis-28398 185 28 our our PRON bracis-28398 185 29 framework framework NOUN bracis-28398 185 30 in in ADP bracis-28398 185 31 a a DET bracis-28398 185 32 data datum NOUN bracis-28398 185 33 sculpting sculpt VERB bracis-28398 185 34 strategy strategy NOUN bracis-28398 185 35 . . PUNCT bracis-28398 186 1 the the DET bracis-28398 186 2 idea idea NOUN bracis-28398 186 3 is be AUX bracis-28398 186 4 to to PART bracis-28398 186 5 remove remove VERB bracis-28398 186 6 the the DET bracis-28398 186 7 instances instance NOUN bracis-28398 186 8 of of ADP bracis-28398 186 9 the the DET bracis-28398 186 10 hardness hardness NOUN bracis-28398 186 11 footprint footprint NOUN bracis-28398 186 12 , , PUNCT bracis-28398 186 13 assuming assume VERB bracis-28398 186 14 they they PRON bracis-28398 186 15 are be AUX bracis-28398 186 16 probably probably ADV bracis-28398 186 17 noisy noisy ADJ bracis-28398 186 18 and and CCONJ bracis-28398 186 19 incorrect incorrect ADJ bracis-28398 186 20 . . PUNCT bracis-28398 187 1 the the DET bracis-28398 187 2 results result NOUN bracis-28398 187 3 of of ADP bracis-28398 187 4 this this DET bracis-28398 187 5 procedure procedure NOUN bracis-28398 187 6 are be AUX bracis-28398 187 7 illustrated illustrate VERB bracis-28398 187 8 in in ADP bracis-28398 187 9 fig fig NOUN bracis-28398 187 10 . . PUNCT bracis-28398 187 11     SPACE bracis-28398 188 1 6 6 NUM bracis-28398 188 2 . . PUNCT bracis-28398 188 3 increasing increase VERB bracis-28398 188 4 thresholds threshold NOUN bracis-28398 188 5 of of ADP bracis-28398 188 6 t t PROPN bracis-28398 188 7 are be AUX bracis-28398 188 8 considered consider VERB bracis-28398 188 9 , , PUNCT bracis-28398 188 10 which which PRON bracis-28398 188 11 means mean VERB bracis-28398 188 12 that that SCONJ bracis-28398 188 13 we we PRON bracis-28398 188 14 start start VERB bracis-28398 188 15 removing remove VERB bracis-28398 188 16 instances instance NOUN bracis-28398 188 17 with with ADP bracis-28398 188 18 the the DET bracis-28398 188 19 highest high ADJ bracis-28398 188 20 hardness hardness ADJ bracis-28398 188 21 levels level NOUN bracis-28398 188 22 . . PUNCT bracis-28398 189 1 the the DET bracis-28398 189 2 average average ADJ bracis-28398 189 3 auc auc NOUN bracis-28398 189 4 performance performance NOUN bracis-28398 189 5 of of ADP bracis-28398 189 6 the the DET bracis-28398 189 7 seven seven NUM bracis-28398 189 8 classifiers classifier NOUN bracis-28398 189 9 from from ADP bracis-28398 189 10 the the DET bracis-28398 189 11 pool pool NOUN bracis-28398 189 12 \(\mathcal \(\mathcal ADJ bracis-28398 189 13 a\ a\ NOUN bracis-28398 189 14 ) ) PUNCT bracis-28398 189 15 is be AUX bracis-28398 189 16 shown show VERB bracis-28398 189 17 in in ADP bracis-28398 189 18 the the DET bracis-28398 189 19 y y NOUN bracis-28398 189 20 - - PUNCT bracis-28398 189 21 axis axis NOUN bracis-28398 189 22 , , PUNCT bracis-28398 189 23 with with ADP bracis-28398 189 24 standard standard ADJ bracis-28398 189 25 deviations deviation NOUN bracis-28398 189 26 in in ADP bracis-28398 189 27 gray gray ADJ bracis-28398 189 28 . . PUNCT bracis-28398 190 1 the the DET bracis-28398 190 2 performance performance NOUN bracis-28398 190 3 achieved achieve VERB bracis-28398 190 4 with with ADP bracis-28398 190 5 no no DET bracis-28398 190 6 sculpting sculpting NOUN bracis-28398 190 7 ( ( PUNCT bracis-28398 190 8 keeping keep VERB bracis-28398 190 9 all all DET bracis-28398 190 10 instances instance NOUN bracis-28398 190 11 ) ) PUNCT bracis-28398 190 12 is be AUX bracis-28398 190 13 shown show VERB bracis-28398 190 14 as as ADP bracis-28398 190 15 a a DET bracis-28398 190 16 dotted dotted ADJ bracis-28398 190 17 line line NOUN bracis-28398 190 18 . . PUNCT bracis-28398 191 1 in in ADP bracis-28398 191 2 fig fig NOUN bracis-28398 191 3 . . PUNCT bracis-28398 191 4     SPACE bracis-28398 191 5 6a 6a NOUN bracis-28398 191 6 , , PUNCT bracis-28398 191 7 we we PRON bracis-28398 191 8 train train VERB bracis-28398 191 9 and and CCONJ bracis-28398 191 10 test test VERB bracis-28398 191 11 the the DET bracis-28398 191 12 classifier classifier NOUN bracis-28398 191 13 over over ADP bracis-28398 191 14 the the DET bracis-28398 191 15 same same ADJ bracis-28398 191 16 dataset dataset NOUN bracis-28398 191 17 ( ( PUNCT bracis-28398 191 18 hospital hospital NOUN bracis-28398 191 19 1 1 NUM bracis-28398 191 20 ) ) PUNCT bracis-28398 191 21 . . PUNCT bracis-28398 192 1 as as SCONJ bracis-28398 192 2 expected expect VERB bracis-28398 192 3 , , PUNCT bracis-28398 192 4 removing remove VERB bracis-28398 192 5 the the DET bracis-28398 192 6 hard hard ADJ bracis-28398 192 7 instances instance NOUN bracis-28398 192 8 increases increase VERB bracis-28398 192 9 the the DET bracis-28398 192 10 accuracy accuracy NOUN bracis-28398 192 11 achieved achieve VERB bracis-28398 192 12 . . PUNCT bracis-28398 193 1 in in ADP bracis-28398 193 2 fig fig NOUN bracis-28398 193 3 . . PUNCT bracis-28398 193 4     SPACE bracis-28398 193 5 6b 6b NOUN bracis-28398 193 6 , , PUNCT bracis-28398 193 7 the the DET bracis-28398 193 8 sculpting sculpting NOUN bracis-28398 193 9 takes take VERB bracis-28398 193 10 place place NOUN bracis-28398 193 11 at at ADP bracis-28398 193 12 instances instance NOUN bracis-28398 193 13 from from ADP bracis-28398 193 14 dataset dataset ADJ bracis-28398 193 15 hospital hospital NOUN bracis-28398 193 16 1 1 NUM bracis-28398 193 17 , , PUNCT bracis-28398 193 18 used use VERB bracis-28398 193 19 for for ADP bracis-28398 193 20 training training NOUN bracis-28398 193 21 , , PUNCT bracis-28398 193 22 but but CCONJ bracis-28398 193 23 test test NOUN bracis-28398 193 24 takes take VERB bracis-28398 193 25 place place NOUN bracis-28398 193 26 at at ADP bracis-28398 193 27 a a DET bracis-28398 193 28 different different ADJ bracis-28398 193 29 dataset dataset NOUN bracis-28398 193 30 , , PUNCT bracis-28398 193 31 hospital hospital NOUN bracis-28398 193 32 2 2 NUM bracis-28398 193 33 . . PUNCT bracis-28398 194 1 this this DET bracis-28398 194 2 second second ADJ bracis-28398 194 3 analysis analysis NOUN bracis-28398 194 4 already already ADV bracis-28398 194 5 entails entail VERB bracis-28398 194 6 a a DET bracis-28398 194 7 higher high ADJ bracis-28398 194 8 difficulty difficulty NOUN bracis-28398 194 9 level level NOUN bracis-28398 194 10 , , PUNCT bracis-28398 194 11 since since SCONJ bracis-28398 194 12 it it PRON bracis-28398 194 13 presupposes presuppose VERB bracis-28398 194 14 the the DET bracis-28398 194 15 profiles profile NOUN bracis-28398 194 16 of of ADP bracis-28398 194 17 patients patient NOUN bracis-28398 194 18 from from ADP bracis-28398 194 19 hospital hospital NOUN bracis-28398 194 20 1 1 NUM bracis-28398 194 21 are be AUX bracis-28398 194 22 predictive predictive ADJ bracis-28398 194 23 of of ADP bracis-28398 194 24 the the DET bracis-28398 194 25 conditions condition NOUN bracis-28398 194 26 of of ADP bracis-28398 194 27 the the DET bracis-28398 194 28 patients patient NOUN bracis-28398 194 29 from from ADP bracis-28398 194 30 hospital hospital NOUN bracis-28398 194 31 2 2 NUM bracis-28398 194 32 . . PUNCT bracis-28398 195 1 in in ADP bracis-28398 195 2 this this DET bracis-28398 195 3 case case NOUN bracis-28398 195 4 , , PUNCT bracis-28398 195 5 there there PRON bracis-28398 195 6 is be VERB bracis-28398 195 7 only only ADV bracis-28398 195 8 a a DET bracis-28398 195 9 subtle subtle ADJ bracis-28398 195 10 increase increase NOUN bracis-28398 195 11 of of ADP bracis-28398 195 12 accuracy accuracy NOUN bracis-28398 195 13 over over ADP bracis-28398 195 14 the the DET bracis-28398 195 15 original original ADJ bracis-28398 195 16 baseline baseline NOUN bracis-28398 195 17 ( ( PUNCT bracis-28398 195 18 the the DET bracis-28398 195 19 dotted dotted ADJ bracis-28398 195 20 line line NOUN bracis-28398 195 21 ) ) PUNCT bracis-28398 195 22 , , PUNCT bracis-28398 195 23 for for ADP bracis-28398 195 24 t t PROPN bracis-28398 195 25 around around ADP bracis-28398 195 26 0.30 0.30 NUM bracis-28398 195 27 and and CCONJ bracis-28398 195 28 0.34 0.34 NUM bracis-28398 195 29 . . PUNCT bracis-28398 196 1 but but CCONJ bracis-28398 196 2 the the DET bracis-28398 196 3 results result NOUN bracis-28398 196 4 are be AUX bracis-28398 196 5 detrimental detrimental ADJ bracis-28398 196 6 for for ADP bracis-28398 196 7 other other ADJ bracis-28398 196 8 t t PROPN bracis-28398 196 9 values value NOUN bracis-28398 196 10 . . PUNCT bracis-28398 197 1 that that PRON bracis-28398 197 2 is be AUX bracis-28398 197 3 , , PUNCT bracis-28398 197 4 in in ADP bracis-28398 197 5 this this DET bracis-28398 197 6 case case NOUN bracis-28398 197 7 the the DET bracis-28398 197 8 benefit benefit NOUN bracis-28398 197 9 occurs occur VERB bracis-28398 197 10 only only ADV bracis-28398 197 11 when when SCONJ bracis-28398 197 12 a a DET bracis-28398 197 13 small small ADJ bracis-28398 197 14 fraction fraction NOUN bracis-28398 197 15 of of ADP bracis-28398 197 16 the the DET bracis-28398 197 17 hardest hard ADJ bracis-28398 197 18 training training NOUN bracis-28398 197 19 instances instance NOUN bracis-28398 197 20 are be AUX bracis-28398 197 21 removed remove VERB bracis-28398 197 22 . . PUNCT bracis-28398 198 1 the the DET bracis-28398 198 2 standard standard ADJ bracis-28398 198 3 deviation deviation NOUN bracis-28398 198 4 values value NOUN bracis-28398 198 5 are be AUX bracis-28398 198 6 also also ADV bracis-28398 198 7 usually usually ADV bracis-28398 198 8 high high ADJ bracis-28398 198 9 , , PUNCT bracis-28398 198 10 indicating indicate VERB bracis-28398 198 11 a a DET bracis-28398 198 12 large large ADJ bracis-28398 198 13 variation variation NOUN bracis-28398 198 14 of of ADP bracis-28398 198 15 results result NOUN bracis-28398 198 16 . . PUNCT bracis-28398 199 1 as as ADP bracis-28398 199 2 in in ADP bracis-28398 199 3     SPACE bracis-28398 199 4 [ [ X bracis-28398 199 5 6 6 NUM bracis-28398 199 6 ] ] PUNCT bracis-28398 199 7 when when SCONJ bracis-28398 199 8 randomly randomly ADV bracis-28398 199 9 removing remove VERB bracis-28398 199 10 ambiguous ambiguous ADJ bracis-28398 199 11 instances instance NOUN bracis-28398 199 12 , , PUNCT bracis-28398 199 13 removing remove VERB bracis-28398 199 14 hard hard ADJ bracis-28398 199 15 instances instance NOUN bracis-28398 199 16 in in ADP bracis-28398 199 17 a a DET bracis-28398 199 18 dataset dataset NOUN bracis-28398 199 19 always always ADV bracis-28398 199 20 increases increase VERB bracis-28398 199 21 model model NOUN bracis-28398 199 22 performance performance NOUN bracis-28398 199 23 for for ADP bracis-28398 199 24 the the DET bracis-28398 199 25 same same ADJ bracis-28398 199 26 dataset dataset NOUN bracis-28398 199 27 ( ( PUNCT bracis-28398 199 28 fig fig NOUN bracis-28398 199 29 . . PUNCT bracis-28398 199 30     SPACE bracis-28398 199 31 6a 6a NOUN bracis-28398 199 32 ) ) PUNCT bracis-28398 199 33 . . PUNCT bracis-28398 200 1 however however ADV bracis-28398 200 2 , , PUNCT bracis-28398 200 3 inside inside ADP bracis-28398 200 4 health health NOUN bracis-28398 200 5 scenarios scenario NOUN bracis-28398 200 6 the the DET bracis-28398 200 7 removal removal NOUN bracis-28398 200 8 of of ADP bracis-28398 200 9 hard hard ADJ bracis-28398 200 10 instances instance NOUN bracis-28398 200 11 can can AUX bracis-28398 200 12 be be AUX bracis-28398 200 13 dangerous dangerous ADJ bracis-28398 200 14 since since SCONJ bracis-28398 200 15 they they PRON bracis-28398 200 16 are be AUX bracis-28398 200 17 not not PART bracis-28398 200 18 necessarily necessarily ADV bracis-28398 200 19 noisy noisy ADJ bracis-28398 200 20 or or CCONJ bracis-28398 200 21 incorrect incorrect ADJ bracis-28398 200 22 . . PUNCT bracis-28398 201 1 instead instead ADV bracis-28398 201 2 they they PRON bracis-28398 201 3 can can AUX bracis-28398 201 4 represent represent VERB bracis-28398 201 5 sub sub NOUN bracis-28398 201 6 populations population NOUN bracis-28398 201 7 with with ADP bracis-28398 201 8 atypical atypical ADJ bracis-28398 201 9 yet yet CCONJ bracis-28398 201 10 correct correct ADJ bracis-28398 201 11 features feature NOUN bracis-28398 201 12 ’ ' PUNCT bracis-28398 201 13 values value NOUN bracis-28398 201 14 . . PUNCT bracis-28398 202 1 so so ADV bracis-28398 202 2 the the DET bracis-28398 202 3 models model NOUN bracis-28398 202 4 might might AUX bracis-28398 202 5 become become VERB bracis-28398 202 6 incapable incapable ADJ bracis-28398 202 7 to to PART bracis-28398 202 8 correctly correctly ADV bracis-28398 202 9 classify classify VERB bracis-28398 202 10 these these DET bracis-28398 202 11 types type NOUN bracis-28398 202 12 of of ADP bracis-28398 202 13 instances instance NOUN bracis-28398 202 14 . . PUNCT bracis-28398 203 1 this this PRON bracis-28398 203 2 might might AUX bracis-28398 203 3 explain explain VERB bracis-28398 203 4 why why SCONJ bracis-28398 203 5 when when SCONJ bracis-28398 203 6 sculpting sculpt VERB bracis-28398 203 7 data datum NOUN bracis-28398 203 8 from from ADP bracis-28398 203 9 hospital hospital NOUN bracis-28398 203 10 1 1 NUM bracis-28398 203 11 and and CCONJ bracis-28398 203 12 testing test VERB bracis-28398 203 13 the the DET bracis-28398 203 14 results result NOUN bracis-28398 203 15 in in ADP bracis-28398 203 16 data datum NOUN bracis-28398 203 17 from from ADP bracis-28398 203 18 hospital hospital NOUN bracis-28398 203 19 2 2 NUM bracis-28398 203 20 the the DET bracis-28398 203 21 accuracy accuracy NOUN bracis-28398 203 22 decreases decrease VERB bracis-28398 203 23 after after ADP bracis-28398 203 24 a a DET bracis-28398 203 25 threshold threshold NOUN bracis-28398 203 26 around around ADP bracis-28398 203 27 0.34 0.34 NUM bracis-28398 203 28 . . PUNCT bracis-28398 204 1 therefore therefore ADV bracis-28398 204 2 , , PUNCT bracis-28398 204 3 a a DET bracis-28398 204 4 more more ADV bracis-28398 204 5 informed informed ADJ bracis-28398 204 6 data datum NOUN bracis-28398 204 7 sculpting sculpting NOUN bracis-28398 204 8 process process NOUN bracis-28398 204 9 is be AUX bracis-28398 204 10 needed need VERB bracis-28398 204 11 . . PUNCT bracis-28398 205 1 4.3 4.3 NUM bracis-28398 205 2 analyzing analyze VERB bracis-28398 205 3 groups group NOUN bracis-28398 205 4 of of ADP bracis-28398 205 5     SPACE bracis-28398 205 6 data datum NOUN bracis-28398 205 7 with with ADP bracis-28398 205 8     SPACE bracis-28398 205 9 a a DET bracis-28398 205 10     SPACE bracis-28398 205 11 domain domain NOUN bracis-28398 205 12 expert expert NOUN bracis-28398 205 13 here here ADV bracis-28398 205 14 we we PRON bracis-28398 205 15 inspect inspect VERB bracis-28398 205 16 closely closely ADJ bracis-28398 205 17 groups group NOUN bracis-28398 205 18 of of ADP bracis-28398 205 19 hard hard ADJ bracis-28398 205 20 instances instance NOUN bracis-28398 205 21 in in ADP bracis-28398 205 22 a a DET bracis-28398 205 23 dataset dataset NOUN bracis-28398 205 24 and and CCONJ bracis-28398 205 25 show show VERB bracis-28398 205 26 that that SCONJ bracis-28398 205 27 more more ADJ bracis-28398 205 28 guided guide VERB bracis-28398 205 29 insights insight NOUN bracis-28398 205 30 can can AUX bracis-28398 205 31 be be AUX bracis-28398 205 32 devised devise VERB bracis-28398 205 33 . . PUNCT bracis-28398 206 1 this this PRON bracis-28398 206 2 is be AUX bracis-28398 206 3 possible possible ADJ bracis-28398 206 4 taking take VERB bracis-28398 206 5 advantage advantage NOUN bracis-28398 206 6 of of ADP bracis-28398 206 7 the the DET bracis-28398 206 8 different different ADJ bracis-28398 206 9 possibilities possibility NOUN bracis-28398 206 10 of of ADP bracis-28398 206 11 visual visual ADJ bracis-28398 206 12 inspection inspection NOUN bracis-28398 206 13 of of ADP bracis-28398 206 14 the the DET bracis-28398 206 15 hardness hardness NOUN bracis-28398 206 16 embedding embedding NOUN bracis-28398 206 17 , , PUNCT bracis-28398 206 18 which which PRON bracis-28398 206 19 can can AUX bracis-28398 206 20 have have VERB bracis-28398 206 21 the the DET bracis-28398 206 22 instances instance NOUN bracis-28398 206 23 colored color VERB bracis-28398 206 24 according accord VERB bracis-28398 206 25 to to ADP bracis-28398 206 26 all all DET bracis-28398 206 27 features feature NOUN bracis-28398 206 28 , , PUNCT bracis-28398 206 29 meta meta X bracis-28398 206 30 - - PUNCT bracis-28398 206 31 features feature NOUN bracis-28398 206 32 and and CCONJ bracis-28398 206 33 algorithmic algorithmic ADJ bracis-28398 206 34 performance performance NOUN bracis-28398 206 35 values value NOUN bracis-28398 206 36 . . PUNCT bracis-28398 207 1 in in ADP bracis-28398 207 2 spite spite NOUN bracis-28398 207 3 of of ADP bracis-28398 207 4 the the DET bracis-28398 207 5 easy easy ADJ bracis-28398 207 6 profile profile NOUN bracis-28398 207 7 of of ADP bracis-28398 207 8 the the DET bracis-28398 207 9 severity severity NOUN bracis-28398 207 10 dataset dataset NOUN bracis-28398 207 11 ( ( PUNCT bracis-28398 207 12 fig fig NOUN bracis-28398 207 13 . . PUNCT bracis-28398 207 14     SPACE bracis-28398 207 15 2a 2a NUM bracis-28398 207 16 ) ) PUNCT bracis-28398 207 17 , , PUNCT bracis-28398 207 18 there there PRON bracis-28398 207 19 are be VERB bracis-28398 207 20 still still ADV bracis-28398 207 21 instances instance NOUN bracis-28398 207 22 with with ADP bracis-28398 207 23 a a DET bracis-28398 207 24 high high ADJ bracis-28398 207 25 level level NOUN bracis-28398 207 26 of of ADP bracis-28398 207 27 difficulty difficulty NOUN bracis-28398 207 28 placed place VERB bracis-28398 207 29 in in ADP bracis-28398 207 30 the the DET bracis-28398 207 31 upper upper ADJ bracis-28398 207 32 left left ADJ bracis-28398 207 33 corner corner NOUN bracis-28398 207 34 of of ADP bracis-28398 207 35 the the DET bracis-28398 207 36 hardness hardness NOUN bracis-28398 207 37 embedding embed VERB bracis-28398 207 38 . . PUNCT bracis-28398 208 1 most most ADJ bracis-28398 208 2 of of ADP bracis-28398 208 3 the the DET bracis-28398 208 4 patients patient NOUN bracis-28398 208 5 with with ADP bracis-28398 208 6 a a DET bracis-28398 208 7 severe severe ADJ bracis-28398 208 8 condition condition NOUN bracis-28398 208 9 are be AUX bracis-28398 208 10 placed place VERB bracis-28398 208 11 in in ADP bracis-28398 208 12 the the DET bracis-28398 208 13 bottom bottom NOUN bracis-28398 208 14 of of ADP bracis-28398 208 15 the the DET bracis-28398 208 16 space space NOUN bracis-28398 208 17 , , PUNCT bracis-28398 208 18 indicating indicate VERB bracis-28398 208 19 this this DET bracis-28398 208 20 class class NOUN bracis-28398 208 21 is be AUX bracis-28398 208 22 fairly fairly ADV bracis-28398 208 23 easy easy ADJ bracis-28398 208 24 to to PART bracis-28398 208 25 classify classify VERB bracis-28398 208 26 ( ( PUNCT bracis-28398 208 27 group group NOUN bracis-28398 208 28 1 1 NUM bracis-28398 208 29 ) ) PUNCT bracis-28398 208 30 . . PUNCT bracis-28398 209 1 in in ADP bracis-28398 209 2 the the DET bracis-28398 209 3 top top ADJ bracis-28398 209 4 left left NOUN bracis-28398 209 5 of of ADP bracis-28398 209 6 the the DET bracis-28398 209 7 space space NOUN bracis-28398 209 8 there there PRON bracis-28398 209 9 are be VERB bracis-28398 209 10 non non ADJ bracis-28398 209 11 - - ADJ bracis-28398 209 12 severe severe ADJ bracis-28398 209 13 patients patient NOUN bracis-28398 209 14 very very ADV bracis-28398 209 15 hard hard ADV bracis-28398 209 16 to to PART bracis-28398 209 17 classify classify VERB bracis-28398 209 18 ( ( PUNCT bracis-28398 209 19 group group NOUN bracis-28398 209 20 2 2 NUM bracis-28398 209 21 ) ) PUNCT bracis-28398 209 22 and and CCONJ bracis-28398 209 23 also also ADV bracis-28398 209 24 some some DET bracis-28398 209 25 severe severe ADJ bracis-28398 209 26 cases case NOUN bracis-28398 209 27 hard hard ADJ bracis-28398 209 28 to to PART bracis-28398 209 29 classify classify VERB bracis-28398 209 30 in in ADP bracis-28398 209 31 the the DET bracis-28398 209 32 top top NOUN bracis-28398 209 33 of of ADP bracis-28398 209 34 the the DET bracis-28398 209 35 hardness hardness NOUN bracis-28398 209 36 embedding embed VERB bracis-28398 209 37 ( ( PUNCT bracis-28398 209 38 group group NOUN bracis-28398 209 39 3 3 NUM bracis-28398 209 40 ) ) PUNCT bracis-28398 209 41 . . PUNCT bracis-28398 210 1 patients patient NOUN bracis-28398 210 2 without without ADP bracis-28398 210 3 a a DET bracis-28398 210 4 severe severe ADJ bracis-28398 210 5 condition condition NOUN bracis-28398 210 6 are be AUX bracis-28398 210 7 mostly mostly ADV bracis-28398 210 8 placed place VERB bracis-28398 210 9 in in ADP bracis-28398 210 10 the the DET bracis-28398 210 11 center center NOUN bracis-28398 210 12 of of ADP bracis-28398 210 13 the the DET bracis-28398 210 14 space space NOUN bracis-28398 210 15 ( ( PUNCT bracis-28398 210 16 group group NOUN bracis-28398 210 17 4 4 NUM bracis-28398 210 18 ) ) PUNCT bracis-28398 210 19 , , PUNCT bracis-28398 210 20 being be AUX bracis-28398 210 21 also also ADV bracis-28398 210 22 easy easy ADJ bracis-28398 210 23 to to PART bracis-28398 210 24 classify classify VERB bracis-28398 210 25 . . PUNCT bracis-28398 211 1 patients patient NOUN bracis-28398 211 2 composing compose VERB bracis-28398 211 3 group group NOUN bracis-28398 211 4 2 2 NUM bracis-28398 211 5 present present NOUN bracis-28398 211 6 at at ADV bracis-28398 211 7 least least ADJ bracis-28398 211 8 one one NUM bracis-28398 211 9 of of ADP bracis-28398 211 10 three three NUM bracis-28398 211 11 attributes attribute NOUN bracis-28398 211 12 heavily heavily ADV bracis-28398 211 13 correlated correlate VERB bracis-28398 211 14 with with ADP bracis-28398 211 15 a a DET bracis-28398 211 16 severe severe ADJ bracis-28398 211 17 condition condition NOUN bracis-28398 211 18 . . PUNCT bracis-28398 212 1 as as ADP bracis-28398 212 2 a a DET bracis-28398 212 3 consequence consequence NOUN bracis-28398 212 4 , , PUNCT bracis-28398 212 5 they they PRON bracis-28398 212 6 have have VERB bracis-28398 212 7 a a DET bracis-28398 212 8 low low ADJ bracis-28398 212 9 likelihood likelihood NOUN bracis-28398 212 10 of of ADP bracis-28398 212 11 belonging belong VERB bracis-28398 212 12 to to ADP bracis-28398 212 13 their their PRON bracis-28398 212 14 registered register VERB bracis-28398 212 15 class class NOUN bracis-28398 212 16 ( ( PUNCT bracis-28398 212 17 high high ADJ bracis-28398 212 18 cl cl NOUN bracis-28398 212 19 values value NOUN bracis-28398 212 20 ) ) PUNCT bracis-28398 212 21 . . PUNCT bracis-28398 213 1 to to PART bracis-28398 213 2 understand understand VERB bracis-28398 213 3 why why SCONJ bracis-28398 213 4 these these DET bracis-28398 213 5 patients patient NOUN bracis-28398 213 6 did do AUX bracis-28398 213 7 not not PART bracis-28398 213 8 require require VERB bracis-28398 213 9 an an DET bracis-28398 213 10 extended extended ADJ bracis-28398 213 11 hospitalization hospitalization NOUN bracis-28398 213 12 or or CCONJ bracis-28398 213 13 did do AUX bracis-28398 213 14 not not PART bracis-28398 213 15 evolve evolve VERB bracis-28398 213 16 to to ADP bracis-28398 213 17 death death NOUN bracis-28398 213 18 , , PUNCT bracis-28398 213 19 we we PRON bracis-28398 213 20 inspected inspect VERB bracis-28398 213 21 the the DET bracis-28398 213 22 raw raw ADJ bracis-28398 213 23 databases database NOUN bracis-28398 213 24 , , PUNCT bracis-28398 213 25 obtaining obtain VERB bracis-28398 213 26 some some DET bracis-28398 213 27 valuable valuable ADJ bracis-28398 213 28 insights insight NOUN bracis-28398 213 29 . . PUNCT bracis-28398 214 1 from from ADP bracis-28398 214 2 the the DET bracis-28398 214 3 90 90 NUM bracis-28398 214 4 instances instance NOUN bracis-28398 214 5 , , PUNCT bracis-28398 214 6 17 17 NUM bracis-28398 214 7 were be AUX bracis-28398 214 8 mislabeled mislabele VERB bracis-28398 214 9 due due ADP bracis-28398 214 10 to to ADP bracis-28398 214 11 data datum NOUN bracis-28398 214 12 preprocessing preprocessing NOUN bracis-28398 214 13 failures failure NOUN bracis-28398 214 14 and and CCONJ bracis-28398 214 15 should should AUX bracis-28398 214 16 be be AUX bracis-28398 214 17 corrected correct VERB bracis-28398 214 18 . . PUNCT bracis-28398 215 1 the the DET bracis-28398 215 2 remaining remain VERB bracis-28398 215 3 were be AUX bracis-28398 215 4 hospitalized hospitalize VERB bracis-28398 215 5 but but CCONJ bracis-28398 215 6 released release VERB bracis-28398 215 7 before before ADP bracis-28398 215 8 ten ten NUM bracis-28398 215 9 days day NOUN bracis-28398 215 10 of of ADP bracis-28398 215 11 hospitalization hospitalization NOUN bracis-28398 215 12 . . PUNCT bracis-28398 216 1 between between ADP bracis-28398 216 2 then then ADV bracis-28398 216 3 , , PUNCT bracis-28398 216 4 nine nine NUM bracis-28398 216 5 patients patient NOUN bracis-28398 216 6 were be AUX bracis-28398 216 7 under under ADP bracis-28398 216 8 35 35 NUM bracis-28398 216 9 years year NOUN bracis-28398 216 10 old old ADJ bracis-28398 216 11 , , PUNCT bracis-28398 216 12 which which PRON bracis-28398 216 13 can can AUX bracis-28398 216 14 explain explain VERB bracis-28398 216 15 the the DET bracis-28398 216 16 quick quick ADJ bracis-28398 216 17 recovery recovery NOUN bracis-28398 216 18 . . PUNCT bracis-28398 217 1 for for ADP bracis-28398 217 2 others other NOUN bracis-28398 217 3 , , PUNCT bracis-28398 217 4 the the DET bracis-28398 217 5 short short ADJ bracis-28398 217 6 hospitalization hospitalization NOUN bracis-28398 217 7 could could AUX bracis-28398 217 8 be be AUX bracis-28398 217 9 explained explain VERB bracis-28398 217 10 either either CCONJ bracis-28398 217 11 because because SCONJ bracis-28398 217 12 of of ADP bracis-28398 217 13 disease disease NOUN bracis-28398 217 14 recovery recovery NOUN bracis-28398 217 15 despite despite SCONJ bracis-28398 217 16 the the DET bracis-28398 217 17 apparent apparent ADJ bracis-28398 217 18 severity severity NOUN bracis-28398 217 19 of of ADP bracis-28398 217 20 the the DET bracis-28398 217 21 case case NOUN bracis-28398 217 22 , , PUNCT bracis-28398 217 23 or or CCONJ bracis-28398 217 24 by by ADP bracis-28398 217 25 an an DET bracis-28398 217 26 external external ADJ bracis-28398 217 27 factor factor NOUN bracis-28398 217 28 , , PUNCT bracis-28398 217 29 like like ADP bracis-28398 217 30 reduced reduce VERB bracis-28398 217 31 hospital hospital NOUN bracis-28398 217 32 capacity capacity NOUN bracis-28398 217 33 for for ADP bracis-28398 217 34 receiving receive VERB bracis-28398 217 35 such such ADJ bracis-28398 217 36 patients patient NOUN bracis-28398 217 37 . . PUNCT bracis-28398 218 1 the the DET bracis-28398 218 2 complete complete ADJ bracis-28398 218 3 analysis analysis NOUN bracis-28398 218 4 of of ADP bracis-28398 218 5 the the DET bracis-28398 218 6 other other ADJ bracis-28398 218 7 groups group NOUN bracis-28398 218 8 can can AUX bracis-28398 218 9 be be AUX bracis-28398 218 10 found find VERB bracis-28398 218 11 in in ADP bracis-28398 218 12 our our PRON bracis-28398 218 13 repository repository NOUN bracis-28398 218 14 . . PUNCT bracis-28398 219 1 in in ADP bracis-28398 219 2 the the DET bracis-28398 219 3 hospital hospital NOUN bracis-28398 219 4 1 1 NUM bracis-28398 219 5 dataset dataset NOUN bracis-28398 219 6 , , PUNCT bracis-28398 219 7 the the DET bracis-28398 219 8 nature nature NOUN bracis-28398 219 9 of of ADP bracis-28398 219 10 the the DET bracis-28398 219 11 attributes attribute NOUN bracis-28398 219 12 is be AUX bracis-28398 219 13 different different ADJ bracis-28398 219 14 , , PUNCT bracis-28398 219 15 since since SCONJ bracis-28398 219 16 the the DET bracis-28398 219 17 blood blood NOUN bracis-28398 219 18 tests test NOUN bracis-28398 219 19 have have VERB bracis-28398 219 20 distinct distinct ADJ bracis-28398 219 21 reference reference NOUN bracis-28398 219 22 values value NOUN bracis-28398 219 23 according accord VERB bracis-28398 219 24 to to ADP bracis-28398 219 25 sex sex NOUN bracis-28398 219 26 and and CCONJ bracis-28398 219 27 age age NOUN bracis-28398 219 28 and and CCONJ bracis-28398 219 29 can can AUX bracis-28398 219 30 also also ADV bracis-28398 219 31 change change VERB bracis-28398 219 32 in in ADP bracis-28398 219 33 the the DET bracis-28398 219 34 presence presence NOUN bracis-28398 219 35 of of ADP bracis-28398 219 36 other other ADJ bracis-28398 219 37 diseases disease NOUN bracis-28398 219 38 and and CCONJ bracis-28398 219 39 medication medication NOUN bracis-28398 219 40 . . PUNCT bracis-28398 220 1 these these PRON bracis-28398 220 2 are be AUX bracis-28398 220 3 some some DET bracis-28398 220 4 reasons reason NOUN bracis-28398 220 5 why why SCONJ bracis-28398 220 6 the the DET bracis-28398 220 7 predictive predictive ADJ bracis-28398 220 8 models model NOUN bracis-28398 220 9 perform perform VERB bracis-28398 220 10 poorly poorly ADV bracis-28398 220 11 in in ADP bracis-28398 220 12 this this DET bracis-28398 220 13 dataset dataset NOUN bracis-28398 220 14 . . PUNCT bracis-28398 221 1 the the PRON bracis-28398 221 2 is be AUX bracis-28398 221 3 also also ADV bracis-28398 221 4 offers offer VERB bracis-28398 221 5 insights insight NOUN bracis-28398 221 6 which which PRON bracis-28398 221 7 help help VERB bracis-28398 221 8 to to PART bracis-28398 221 9 understand understand VERB bracis-28398 221 10 why why SCONJ bracis-28398 221 11 , , PUNCT bracis-28398 221 12 how how SCONJ bracis-28398 221 13 , , PUNCT bracis-28398 221 14 and and CCONJ bracis-28398 221 15 where where SCONJ bracis-28398 221 16 the the DET bracis-28398 221 17 ml ml PROPN bracis-28398 221 18 models model NOUN bracis-28398 221 19 are be AUX bracis-28398 221 20 failing fail VERB bracis-28398 221 21 . . PUNCT bracis-28398 222 1 comparing compare VERB bracis-28398 222 2 meta meta ADJ bracis-28398 222 3 - - PUNCT bracis-28398 222 4 feature feature NOUN bracis-28398 222 5 values value NOUN bracis-28398 222 6 for for ADP bracis-28398 222 7 the the DET bracis-28398 222 8 severe severe ADJ bracis-28398 222 9 individuals individual NOUN bracis-28398 222 10 hard hard ADJ bracis-28398 222 11 to to PART bracis-28398 222 12 classify classify VERB bracis-28398 222 13 reveals reveal VERB bracis-28398 222 14 six six NUM bracis-28398 222 15 types type NOUN bracis-28398 222 16 of of ADP bracis-28398 222 17 hard hard ADJ bracis-28398 222 18 instances instance NOUN bracis-28398 222 19 . . PUNCT bracis-28398 223 1 one one NUM bracis-28398 223 2 of of ADP bracis-28398 223 3 these these DET bracis-28398 223 4 groups group NOUN bracis-28398 223 5 is be AUX bracis-28398 223 6 composed compose VERB bracis-28398 223 7 of of ADP bracis-28398 223 8 eight eight NUM bracis-28398 223 9 instances instance NOUN bracis-28398 223 10 labeled label VERB bracis-28398 223 11 as as ADP bracis-28398 223 12 non non ADJ bracis-28398 223 13 - - ADJ bracis-28398 223 14 severe severe ADJ bracis-28398 223 15 , , PUNCT bracis-28398 223 16 although although SCONJ bracis-28398 223 17 they they PRON bracis-28398 223 18 have have VERB bracis-28398 223 19 the the DET bracis-28398 223 20 opposite opposite ADJ bracis-28398 223 21 profile profile NOUN bracis-28398 223 22 . . PUNCT bracis-28398 224 1 indeed indeed ADV bracis-28398 224 2 , , PUNCT bracis-28398 224 3 they they PRON bracis-28398 224 4 have have VERB bracis-28398 224 5 a a DET bracis-28398 224 6 low low ADJ bracis-28398 224 7 likelihood likelihood NOUN bracis-28398 224 8 of of ADP bracis-28398 224 9 belonging belong VERB bracis-28398 224 10 to to ADP bracis-28398 224 11 their their PRON bracis-28398 224 12 registered register VERB bracis-28398 224 13 class class NOUN bracis-28398 224 14 ( ( PUNCT bracis-28398 224 15 cl cl NOUN bracis-28398 224 16 ) ) PUNCT bracis-28398 224 17 and and CCONJ bracis-28398 224 18 are be AUX bracis-28398 224 19 surrounded surround VERB bracis-28398 224 20 by by ADP bracis-28398 224 21 elements element NOUN bracis-28398 224 22 from from ADP bracis-28398 224 23 a a DET bracis-28398 224 24 different different ADJ bracis-28398 224 25 class class NOUN bracis-28398 224 26 ( ( PUNCT bracis-28398 224 27 n1 n1 NOUN bracis-28398 224 28 ) ) PUNCT bracis-28398 224 29 . . PUNCT bracis-28398 225 1 each each DET bracis-28398 225 2 individual individual NOUN bracis-28398 225 3 in in ADP bracis-28398 225 4 this this DET bracis-28398 225 5 group group NOUN bracis-28398 225 6 is be AUX bracis-28398 225 7 a a DET bracis-28398 225 8 man man NOUN bracis-28398 225 9 between between ADP bracis-28398 225 10 fifty fifty NUM bracis-28398 225 11 and and CCONJ bracis-28398 225 12 sixty sixty NUM bracis-28398 225 13 years year NOUN bracis-28398 225 14 old old ADJ bracis-28398 225 15 , , PUNCT bracis-28398 225 16 with with ADP bracis-28398 225 17 low low ADJ bracis-28398 225 18 percentage percentage NOUN bracis-28398 225 19 of of ADP bracis-28398 225 20 lymphocytes lymphocyte NOUN bracis-28398 225 21 and and CCONJ bracis-28398 225 22 c c NOUN bracis-28398 225 23 - - PUNCT bracis-28398 225 24 reactive reactive ADJ bracis-28398 225 25 protein protein NOUN bracis-28398 225 26 count count NOUN bracis-28398 225 27 as as ADV bracis-28398 225 28 well well ADV bracis-28398 225 29 as as ADP bracis-28398 225 30 high high ADJ bracis-28398 225 31 percentages percentage NOUN bracis-28398 225 32 of of ADP bracis-28398 225 33 neutrophils neutrophil NOUN bracis-28398 225 34 . . PUNCT bracis-28398 226 1 investigating investigate VERB bracis-28398 226 2 the the DET bracis-28398 226 3 original original ADJ bracis-28398 226 4 database database NOUN bracis-28398 226 5 it it PRON bracis-28398 226 6 is be AUX bracis-28398 226 7 possible possible ADJ bracis-28398 226 8 to to PART bracis-28398 226 9 follow follow VERB bracis-28398 226 10 up up ADP bracis-28398 226 11 the the DET bracis-28398 226 12 values value NOUN bracis-28398 226 13 of of ADP bracis-28398 226 14 these these DET bracis-28398 226 15 same same ADJ bracis-28398 226 16 blood blood NOUN bracis-28398 226 17 tests test NOUN bracis-28398 226 18 in in ADP bracis-28398 226 19 the the DET bracis-28398 226 20 next next ADJ bracis-28398 226 21 few few ADJ bracis-28398 226 22 days day NOUN bracis-28398 226 23 . . PUNCT bracis-28398 227 1 this this DET bracis-28398 227 2 inspection inspection NOUN bracis-28398 227 3 reveals reveal VERB bracis-28398 227 4 that that SCONJ bracis-28398 227 5 , , PUNCT bracis-28398 227 6 besides besides SCONJ bracis-28398 227 7 their their PRON bracis-28398 227 8 initial initial ADJ bracis-28398 227 9 condition condition NOUN bracis-28398 227 10 , , PUNCT bracis-28398 227 11 all all PRON bracis-28398 227 12 of of ADP bracis-28398 227 13 them they PRON bracis-28398 227 14 recovered recover VERB bracis-28398 227 15 before before ADP bracis-28398 227 16 14 14 NUM bracis-28398 227 17 days day NOUN bracis-28398 227 18 of of ADP bracis-28398 227 19 hospitalization hospitalization NOUN bracis-28398 227 20 ( ( PUNCT bracis-28398 227 21 the the DET bracis-28398 227 22 proxy proxy NOUN bracis-28398 227 23 for for ADP bracis-28398 227 24 considering consider VERB bracis-28398 227 25 a a DET bracis-28398 227 26 severe severe ADJ bracis-28398 227 27 patient patient NOUN bracis-28398 227 28 in in ADP bracis-28398 227 29 this this DET bracis-28398 227 30 dataset dataset NOUN bracis-28398 227 31 ) ) PUNCT bracis-28398 227 32 . . PUNCT bracis-28398 228 1 therefore therefore ADV bracis-28398 228 2 , , PUNCT bracis-28398 228 3 this this DET bracis-28398 228 4 group group NOUN bracis-28398 228 5 is be AUX bracis-28398 228 6 composed compose VERB bracis-28398 228 7 of of ADP bracis-28398 228 8 outlier outlier NOUN bracis-28398 228 9 patients patient NOUN bracis-28398 228 10 with with ADP bracis-28398 228 11 a a DET bracis-28398 228 12 faster fast ADJ bracis-28398 228 13 recovery recovery NOUN bracis-28398 228 14 than than SCONJ bracis-28398 228 15 would would AUX bracis-28398 228 16 be be AUX bracis-28398 228 17 initially initially ADV bracis-28398 228 18 expected expect VERB bracis-28398 228 19 . . PUNCT bracis-28398 229 1 another another DET bracis-28398 229 2 group group NOUN bracis-28398 229 3 is be AUX bracis-28398 229 4 composed compose VERB bracis-28398 229 5 of of ADP bracis-28398 229 6 30 30 NUM bracis-28398 229 7 patients patient NOUN bracis-28398 229 8 that that PRON bracis-28398 229 9 also also ADV bracis-28398 229 10 present present VERB bracis-28398 229 11 a a DET bracis-28398 229 12 clinical clinical ADJ bracis-28398 229 13 condition condition NOUN bracis-28398 229 14 similar similar ADJ bracis-28398 229 15 to to ADP bracis-28398 229 16 the the DET bracis-28398 229 17 severe severe ADJ bracis-28398 229 18 class class NOUN bracis-28398 229 19 , , PUNCT bracis-28398 229 20 but but CCONJ bracis-28398 229 21 are be AUX bracis-28398 229 22 labeled label VERB bracis-28398 229 23 as as ADP bracis-28398 229 24 non non ADJ bracis-28398 229 25 - - ADJ bracis-28398 229 26 severe severe ADJ bracis-28398 229 27 patients patient NOUN bracis-28398 229 28 . . PUNCT bracis-28398 230 1 all all DET bracis-28398 230 2 meta meta ADJ bracis-28398 230 3 - - PUNCT bracis-28398 230 4 features feature NOUN bracis-28398 230 5 are be AUX bracis-28398 230 6 high high ADJ bracis-28398 230 7 for for ADP bracis-28398 230 8 these these DET bracis-28398 230 9 patients patient NOUN bracis-28398 230 10 ( ( PUNCT bracis-28398 230 11 cl cl NOUN bracis-28398 230 12 , , PUNCT bracis-28398 230 13 dcp dcp NOUN bracis-28398 230 14 and and CCONJ bracis-28398 230 15 n1 n1 NOUN bracis-28398 230 16 ) ) PUNCT bracis-28398 230 17 , , PUNCT bracis-28398 230 18 revealing reveal VERB bracis-28398 230 19 they they PRON bracis-28398 230 20 are be AUX bracis-28398 230 21 difficult difficult ADJ bracis-28398 230 22 to to PART bracis-28398 230 23 classify classify VERB bracis-28398 230 24 according accord VERB bracis-28398 230 25 to to ADP bracis-28398 230 26 different different ADJ bracis-28398 230 27 perspectives perspective NOUN bracis-28398 230 28 . . PUNCT bracis-28398 231 1 analyzing analyze VERB bracis-28398 231 2 the the DET bracis-28398 231 3 original original ADJ bracis-28398 231 4 database database NOUN bracis-28398 231 5 , , PUNCT bracis-28398 231 6 we we PRON bracis-28398 231 7 found find VERB bracis-28398 231 8 an an DET bracis-28398 231 9 unexpected unexpected ADJ bracis-28398 231 10 pattern pattern NOUN bracis-28398 231 11 . . PUNCT bracis-28398 232 1 some some PRON bracis-28398 232 2 of of ADP bracis-28398 232 3 these these DET bracis-28398 232 4 patients patient NOUN bracis-28398 232 5 are be AUX bracis-28398 232 6 released release VERB bracis-28398 232 7 without without ADP bracis-28398 232 8 recovering recover VERB bracis-28398 232 9 ( ( PUNCT bracis-28398 232 10 still still ADV bracis-28398 232 11 presenting present VERB bracis-28398 232 12 low low ADJ bracis-28398 232 13 percentage percentage NOUN bracis-28398 232 14 of of ADP bracis-28398 232 15 lymphocytes lymphocyte NOUN bracis-28398 232 16 and and CCONJ bracis-28398 232 17 c c NOUN bracis-28398 232 18 - - PUNCT bracis-28398 232 19 reactive reactive ADJ bracis-28398 232 20 protein protein NOUN bracis-28398 232 21 counts count NOUN bracis-28398 232 22 ) ) PUNCT bracis-28398 232 23 and and CCONJ bracis-28398 232 24 sometimes sometimes ADV bracis-28398 232 25 in in ADP bracis-28398 232 26 a a DET bracis-28398 232 27 worse bad ADJ bracis-28398 232 28 condition condition NOUN bracis-28398 232 29 . . PUNCT bracis-28398 233 1 we we PRON bracis-28398 233 2 could could AUX bracis-28398 233 3 not not PART bracis-28398 233 4 recover recover VERB bracis-28398 233 5 the the DET bracis-28398 233 6 real real ADJ bracis-28398 233 7 reason reason NOUN bracis-28398 233 8 for for ADP bracis-28398 233 9 hospital hospital NOUN bracis-28398 233 10 discharge discharge NOUN bracis-28398 233 11 . . PUNCT bracis-28398 234 1 these these DET bracis-28398 234 2 patients patient NOUN bracis-28398 234 3 might might AUX bracis-28398 234 4 have have AUX bracis-28398 234 5 been be AUX bracis-28398 234 6 transferred transfer VERB bracis-28398 234 7 between between ADP bracis-28398 234 8 hospitals hospital NOUN bracis-28398 234 9 , , PUNCT bracis-28398 234 10 although although SCONJ bracis-28398 234 11 being be AUX bracis-28398 234 12 registered register VERB bracis-28398 234 13 as as ADP bracis-28398 234 14 medical medical ADJ bracis-28398 234 15 releases release NOUN bracis-28398 234 16 in in ADP bracis-28398 234 17 the the DET bracis-28398 234 18 system system NOUN bracis-28398 234 19 . . PUNCT bracis-28398 235 1 these these PRON bracis-28398 235 2 are be AUX bracis-28398 235 3 more more ADV bracis-28398 235 4 suitable suitable ADJ bracis-28398 235 5 candidates candidate NOUN bracis-28398 235 6 for for ADP bracis-28398 235 7 data datum NOUN bracis-28398 235 8 sculpting sculpt VERB bracis-28398 235 9 , , PUNCT bracis-28398 235 10 since since SCONJ bracis-28398 235 11 they they PRON bracis-28398 235 12 might might AUX bracis-28398 235 13 be be AUX bracis-28398 235 14 noisy noisy ADJ bracis-28398 235 15 , , PUNCT bracis-28398 235 16 as as SCONJ bracis-28398 235 17 justified justify VERB bracis-28398 235 18 by by ADP bracis-28398 235 19 the the DET bracis-28398 235 20 high high ADJ bracis-28398 235 21 values value NOUN bracis-28398 235 22 for for ADP bracis-28398 235 23 their their PRON bracis-28398 235 24 meta meta NOUN bracis-28398 235 25 - - PUNCT bracis-28398 235 26 features feature NOUN bracis-28398 235 27 . . PUNCT bracis-28398 236 1 these these DET bracis-28398 236 2 analysis analysis NOUN bracis-28398 236 3 illustrate illustrate VERB bracis-28398 236 4 how how SCONJ bracis-28398 236 5 our our PRON bracis-28398 236 6 framework framework NOUN bracis-28398 236 7 can can AUX bracis-28398 236 8 support support VERB bracis-28398 236 9 the the DET bracis-28398 236 10 insertion insertion NOUN bracis-28398 236 11 of of ADP bracis-28398 236 12 expert expert ADJ bracis-28398 236 13 knowledge knowledge NOUN bracis-28398 236 14 to to PART bracis-28398 236 15 improve improve VERB bracis-28398 236 16 data datum NOUN bracis-28398 236 17 quality quality NOUN bracis-28398 236 18 and and CCONJ bracis-28398 236 19 model model NOUN bracis-28398 236 20 performance performance NOUN bracis-28398 236 21 , , PUNCT bracis-28398 236 22 in in ADP bracis-28398 236 23 a a DET bracis-28398 236 24 more more ADV bracis-28398 236 25 guided guide VERB bracis-28398 236 26 data datum NOUN bracis-28398 236 27 sculpting sculpting NOUN bracis-28398 236 28 process process NOUN bracis-28398 236 29 . . PUNCT bracis-28398 237 1 5 5 NUM bracis-28398 237 2 discussions discussion NOUN bracis-28398 237 3 and and CCONJ bracis-28398 237 4     SPACE bracis-28398 237 5 conclusions conclusion NOUN bracis-28398 237 6 data data NOUN bracis-28398 237 7 - - PUNCT bracis-28398 237 8 centric centric ADJ bracis-28398 237 9 analysis analysis NOUN bracis-28398 237 10 have have AUX bracis-28398 237 11 been be AUX bracis-28398 237 12 taking take VERB bracis-28398 237 13 an an DET bracis-28398 237 14 important important ADJ bracis-28398 237 15 role role NOUN bracis-28398 237 16 in in ADP bracis-28398 237 17 ml ml ADP bracis-28398 237 18 , , PUNCT bracis-28398 237 19 providing provide VERB bracis-28398 237 20 means mean NOUN bracis-28398 237 21 to to PART bracis-28398 237 22 assure assure VERB bracis-28398 237 23 more more ADJ bracis-28398 237 24 trustfulness trustfulness NOUN bracis-28398 237 25 to to ADP bracis-28398 237 26 the the DET bracis-28398 237 27 area area NOUN bracis-28398 237 28 . . PUNCT bracis-28398 238 1 in in ADP bracis-28398 238 2 this this DET bracis-28398 238 3 paper paper NOUN bracis-28398 238 4 , , PUNCT bracis-28398 238 5 we we PRON bracis-28398 238 6 obtain obtain VERB bracis-28398 238 7 interpretations interpretation NOUN bracis-28398 238 8 on on ADP bracis-28398 238 9 reasons reason NOUN bracis-28398 238 10 why why SCONJ bracis-28398 238 11 some some DET bracis-28398 238 12 observations observation NOUN bracis-28398 238 13 of of ADP bracis-28398 238 14 a a DET bracis-28398 238 15 dataset dataset NOUN bracis-28398 238 16 can can AUX bracis-28398 238 17 be be AUX bracis-28398 238 18 considered consider VERB bracis-28398 238 19 hard hard ADJ bracis-28398 238 20 to to PART bracis-28398 238 21 classify classify VERB bracis-28398 238 22 . . PUNCT bracis-28398 239 1 this this PRON bracis-28398 239 2 is be AUX bracis-28398 239 3 done do VERB bracis-28398 239 4 considering consider VERB bracis-28398 239 5 both both CCONJ bracis-28398 239 6 original original ADJ bracis-28398 239 7 input input NOUN bracis-28398 239 8 features feature NOUN bracis-28398 239 9 and and CCONJ bracis-28398 239 10 also also ADV bracis-28398 239 11 meta meta X bracis-28398 239 12 - - PUNCT bracis-28398 239 13 features feature NOUN bracis-28398 239 14 which which PRON bracis-28398 239 15 describe describe VERB bracis-28398 239 16 possible possible ADJ bracis-28398 239 17 abstract abstract ADJ bracis-28398 239 18 structural structural ADJ bracis-28398 239 19 reasons reason NOUN bracis-28398 239 20 explaining explain VERB bracis-28398 239 21 why why SCONJ bracis-28398 239 22 some some DET bracis-28398 239 23 instances instance NOUN bracis-28398 239 24 are be AUX bracis-28398 239 25 hard hard ADJ bracis-28398 239 26 to to PART bracis-28398 239 27 classify classify VERB bracis-28398 239 28 . . PUNCT bracis-28398 240 1 the the DET bracis-28398 240 2 choice choice NOUN bracis-28398 240 3 of of ADP bracis-28398 240 4 the the DET bracis-28398 240 5 instances instance NOUN bracis-28398 240 6 to to PART bracis-28398 240 7 be be AUX bracis-28398 240 8 examined examine VERB bracis-28398 240 9 is be AUX bracis-28398 240 10 based base VERB bracis-28398 240 11 on on ADP bracis-28398 240 12 a a DET bracis-28398 240 13 projection projection NOUN bracis-28398 240 14 of of ADP bracis-28398 240 15 the the DET bracis-28398 240 16 dataset dataset NOUN bracis-28398 240 17 into into ADP bracis-28398 240 18 a a DET bracis-28398 240 19 hardness hardness NOUN bracis-28398 240 20 embedding embed VERB bracis-28398 240 21 showing show VERB bracis-28398 240 22 linear linear ADJ bracis-28398 240 23 trends trend NOUN bracis-28398 240 24 of of ADP bracis-28398 240 25 classification classification NOUN bracis-28398 240 26 difficulty difficulty NOUN bracis-28398 240 27 according accord VERB bracis-28398 240 28 to to ADP bracis-28398 240 29 different different ADJ bracis-28398 240 30 perspectives perspective NOUN bracis-28398 240 31 . . PUNCT bracis-28398 241 1 we we PRON bracis-28398 241 2 show show VERB bracis-28398 241 3 the the DET bracis-28398 241 4 value value NOUN bracis-28398 241 5 of of ADP bracis-28398 241 6 our our PRON bracis-28398 241 7 proposal proposal NOUN bracis-28398 241 8 in in ADP bracis-28398 241 9 gaining gain VERB bracis-28398 241 10 insights insight NOUN bracis-28398 241 11 about about ADP bracis-28398 241 12 data datum NOUN bracis-28398 241 13 from from ADP bracis-28398 241 14 the the DET bracis-28398 241 15 health health NOUN bracis-28398 241 16 domain domain NOUN bracis-28398 241 17 . . PUNCT bracis-28398 242 1 our our PRON bracis-28398 242 2 framework framework NOUN bracis-28398 242 3 was be AUX bracis-28398 242 4 devised devise VERB bracis-28398 242 5 to to ADP bracis-28398 242 6 the the DET bracis-28398 242 7 analysis analysis NOUN bracis-28398 242 8 of of ADP bracis-28398 242 9 datasets dataset NOUN bracis-28398 242 10 in in ADP bracis-28398 242 11 a a DET bracis-28398 242 12 tabular tabular NOUN bracis-28398 242 13 format format NOUN bracis-28398 242 14 , , PUNCT bracis-28398 242 15 which which PRON bracis-28398 242 16 are be AUX bracis-28398 242 17 very very ADV bracis-28398 242 18 abundant abundant ADJ bracis-28398 242 19 in in ADP bracis-28398 242 20 the the DET bracis-28398 242 21 health health NOUN bracis-28398 242 22 domain domain NOUN bracis-28398 242 23 . . PUNCT bracis-28398 243 1 but but CCONJ bracis-28398 243 2 plenty plenty NOUN bracis-28398 243 3 of of ADP bracis-28398 243 4 non non ADJ bracis-28398 243 5 - - ADJ bracis-28398 243 6 structured structured ADJ bracis-28398 243 7 data datum NOUN bracis-28398 243 8 , , PUNCT bracis-28398 243 9 such such ADJ bracis-28398 243 10 as as ADP bracis-28398 243 11 image image NOUN bracis-28398 243 12 and and CCONJ bracis-28398 243 13 text text NOUN bracis-28398 243 14 , , PUNCT bracis-28398 243 15 are be AUX bracis-28398 243 16 also also ADV bracis-28398 243 17 gathered gather VERB bracis-28398 243 18 in in ADP bracis-28398 243 19 this this DET bracis-28398 243 20 domain domain NOUN bracis-28398 243 21 ( ( PUNCT bracis-28398 243 22 eg eg NOUN bracis-28398 243 23 . . PUNCT bracis-28398 244 1 x x X bracis-28398 244 2 - - NOUN bracis-28398 244 3 ray ray NOUN bracis-28398 244 4 images image NOUN bracis-28398 244 5 ) ) PUNCT bracis-28398 244 6 . . PUNCT bracis-28398 245 1 this this PRON bracis-28398 245 2 does do AUX bracis-28398 245 3 not not PART bracis-28398 245 4 prevent prevent VERB bracis-28398 245 5 applying apply VERB bracis-28398 245 6 our our PRON bracis-28398 245 7 framework framework NOUN bracis-28398 245 8 to to ADP bracis-28398 245 9 non non ADJ bracis-28398 245 10 - - ADJ bracis-28398 245 11 structured structured ADJ bracis-28398 245 12 data datum NOUN bracis-28398 245 13 . . PUNCT bracis-28398 246 1 for for ADP bracis-28398 246 2 instance instance NOUN bracis-28398 246 3 , , PUNCT bracis-28398 246 4 it it PRON bracis-28398 246 5 is be AUX bracis-28398 246 6 possible possible ADJ bracis-28398 246 7 to to PART bracis-28398 246 8 extract extract VERB bracis-28398 246 9 structured structured ADJ bracis-28398 246 10 representations representation NOUN bracis-28398 246 11 from from ADP bracis-28398 246 12 non non ADJ bracis-28398 246 13 - - ADJ bracis-28398 246 14 structured structured ADJ bracis-28398 246 15 data datum NOUN bracis-28398 246 16 by by ADP bracis-28398 246 17 imputing impute VERB bracis-28398 246 18 them they PRON bracis-28398 246 19 to to ADP bracis-28398 246 20 some some DET bracis-28398 246 21 deep deep ADJ bracis-28398 246 22 learning learn VERB bracis-28398 246 23 trained train VERB bracis-28398 246 24 models model NOUN bracis-28398 246 25     SPACE bracis-28398 247 1 [ [ X bracis-28398 247 2 26 26 NUM bracis-28398 247 3 ] ] PUNCT bracis-28398 247 4 . . PUNCT bracis-28398 248 1 for for ADP bracis-28398 248 2 obtaining obtain VERB bracis-28398 248 3 the the DET bracis-28398 248 4 hardness hardness NOUN bracis-28398 248 5 embedding embed VERB bracis-28398 248 6 , , PUNCT bracis-28398 248 7 many many ADJ bracis-28398 248 8 meta meta NOUN bracis-28398 248 9 - - PUNCT bracis-28398 248 10 features feature NOUN bracis-28398 248 11 and and CCONJ bracis-28398 248 12 algorithmic algorithmic ADJ bracis-28398 248 13 performance performance NOUN bracis-28398 248 14 measures measure NOUN bracis-28398 248 15 must must AUX bracis-28398 248 16 be be AUX bracis-28398 248 17 extracted extract VERB bracis-28398 248 18 from from ADP bracis-28398 248 19 the the DET bracis-28398 248 20 dataset dataset NOUN bracis-28398 248 21 first first ADV bracis-28398 248 22 . . PUNCT bracis-28398 249 1 this this PRON bracis-28398 249 2 clearly clearly ADV bracis-28398 249 3 implies imply VERB bracis-28398 249 4 in in ADP bracis-28398 249 5 a a DET bracis-28398 249 6 computational computational ADJ bracis-28398 249 7 cost cost NOUN bracis-28398 249 8 which which PRON bracis-28398 249 9 can can AUX bracis-28398 249 10 not not PART bracis-28398 249 11 be be AUX bracis-28398 249 12 disregarded disregard VERB bracis-28398 249 13 . . PUNCT bracis-28398 250 1 running run VERB bracis-28398 250 2 the the DET bracis-28398 250 3 analysis analysis NOUN bracis-28398 250 4 on on ADP bracis-28398 250 5 a a DET bracis-28398 250 6 macbook macbook NOUN bracis-28398 250 7 pro pro X bracis-28398 250 8 os os NOUN bracis-28398 250 9 computer computer NOUN bracis-28398 250 10 ( ( PUNCT bracis-28398 250 11 m1 m1 PROPN bracis-28398 250 12 processor processor NOUN bracis-28398 250 13 with with ADP bracis-28398 250 14 8 8 NUM bracis-28398 250 15 gb gb NOUN bracis-28398 250 16 of of ADP bracis-28398 250 17 memory memory NOUN bracis-28398 250 18 ) ) PUNCT bracis-28398 250 19 , , PUNCT bracis-28398 250 20 the the DET bracis-28398 250 21 average average ADJ bracis-28398 250 22 time time NOUN bracis-28398 250 23 taken take VERB bracis-28398 250 24 to to PART bracis-28398 250 25 obtain obtain VERB bracis-28398 250 26 the the DET bracis-28398 250 27 hardness hardness ADJ bracis-28398 250 28 embeddings embedding NOUN bracis-28398 250 29 for for ADP bracis-28398 250 30 our our PRON bracis-28398 250 31 datasets dataset NOUN bracis-28398 250 32 was be AUX bracis-28398 250 33 around around ADV bracis-28398 250 34 four four NUM bracis-28398 250 35 minutes minute NOUN bracis-28398 250 36 . . PUNCT bracis-28398 251 1 in in ADP bracis-28398 251 2 general general ADJ bracis-28398 251 3 , , PUNCT bracis-28398 251 4 the the DET bracis-28398 251 5 hardness hardness NOUN bracis-28398 251 6 meta meta X bracis-28398 251 7 - - PUNCT bracis-28398 251 8 features feature NOUN bracis-28398 251 9 have have VERB bracis-28398 251 10 at at ADP bracis-28398 251 11 most most ADJ bracis-28398 251 12 a a DET bracis-28398 251 13 quadratic quadratic ADJ bracis-28398 251 14 asymptotic asymptotic ADJ bracis-28398 251 15 cost cost NOUN bracis-28398 251 16 on on ADP bracis-28398 251 17 the the DET bracis-28398 251 18 number number NOUN bracis-28398 251 19 of of ADP bracis-28398 251 20 observations observation NOUN bracis-28398 251 21 the the DET bracis-28398 251 22 dataset dataset NOUN bracis-28398 251 23 has have VERB bracis-28398 251 24 , , PUNCT bracis-28398 251 25 mostly mostly ADV bracis-28398 251 26 because because SCONJ bracis-28398 251 27 some some PRON bracis-28398 251 28 of of ADP bracis-28398 251 29 them they PRON bracis-28398 251 30 require require VERB bracis-28398 251 31 building build VERB bracis-28398 251 32 a a DET bracis-28398 251 33 distance distance NOUN bracis-28398 251 34 matrix matrix NOUN bracis-28398 251 35 between between ADP bracis-28398 251 36 all all DET bracis-28398 251 37 pairs pair NOUN bracis-28398 251 38 of of ADP bracis-28398 251 39 observations observation NOUN bracis-28398 251 40 of of ADP bracis-28398 251 41 the the DET bracis-28398 251 42 dataset dataset NOUN bracis-28398 251 43 . . PUNCT bracis-28398 252 1 but but CCONJ bracis-28398 252 2 the the DET bracis-28398 252 3 main main ADJ bracis-28398 252 4 cost cost NOUN bracis-28398 252 5 is be AUX bracis-28398 252 6 usually usually ADV bracis-28398 252 7 incurred incur VERB bracis-28398 252 8 by by ADP bracis-28398 252 9 the the DET bracis-28398 252 10 cross cross ADJ bracis-28398 252 11 - - ADJ bracis-28398 252 12 validation validation ADJ bracis-28398 252 13 training training NOUN bracis-28398 252 14 - - PUNCT bracis-28398 252 15 testing testing NOUN bracis-28398 252 16 of of ADP bracis-28398 252 17 the the DET bracis-28398 252 18 classifiers classifier NOUN bracis-28398 252 19 in in ADP bracis-28398 252 20 the the DET bracis-28398 252 21 pool pool NOUN bracis-28398 252 22 \(\mathcal \(\mathcal ADJ bracis-28398 252 23 a\ a\ NOUN bracis-28398 252 24 ) ) PUNCT bracis-28398 252 25 . . PUNCT bracis-28398 253 1 computational computational ADJ bracis-28398 253 2 cost cost NOUN bracis-28398 253 3 can can AUX bracis-28398 253 4 be be AUX bracis-28398 253 5 saved save VERB bracis-28398 253 6 by by ADP bracis-28398 253 7 reducing reduce VERB bracis-28398 253 8 the the DET bracis-28398 253 9 amount amount NOUN bracis-28398 253 10 of of ADP bracis-28398 253 11 classifiers classifier NOUN bracis-28398 253 12 considered consider VERB bracis-28398 253 13 , , PUNCT bracis-28398 253 14 although although SCONJ bracis-28398 253 15 we we PRON bracis-28398 253 16 can can AUX bracis-28398 253 17 argue argue VERB bracis-28398 253 18 that that SCONJ bracis-28398 253 19 the the DET bracis-28398 253 20 classifiers classifier NOUN bracis-28398 253 21 chosen choose VERB bracis-28398 253 22 here here ADV bracis-28398 253 23 are be AUX bracis-28398 253 24 common common ADJ bracis-28398 253 25 representatives representative NOUN bracis-28398 253 26 everyone everyone PRON bracis-28398 253 27 tests test VERB bracis-28398 253 28 when when SCONJ bracis-28398 253 29 tabular tabular PROPN bracis-28398 253 30 data datum NOUN bracis-28398 253 31 are be AUX bracis-28398 253 32 concerned concern VERB bracis-28398 253 33 , , PUNCT bracis-28398 253 34 at at ADV bracis-28398 253 35 least least ADJ bracis-28398 253 36 as as ADP bracis-28398 253 37 baselines baseline NOUN bracis-28398 253 38 . . PUNCT bracis-28398 254 1 another another DET bracis-28398 254 2 limitation limitation NOUN bracis-28398 254 3 of of ADP bracis-28398 254 4 our our PRON bracis-28398 254 5 current current ADJ bracis-28398 254 6 approach approach NOUN bracis-28398 254 7 is be AUX bracis-28398 254 8 that that SCONJ bracis-28398 254 9 data datum NOUN bracis-28398 254 10 must must AUX bracis-28398 254 11 be be AUX bracis-28398 254 12 labeled label VERB bracis-28398 254 13 in in ADP bracis-28398 254 14 advance advance NOUN bracis-28398 254 15 . . PUNCT bracis-28398 255 1 we we PRON bracis-28398 255 2 shall shall AUX bracis-28398 255 3 investigate investigate VERB bracis-28398 255 4 strategies strategy NOUN bracis-28398 255 5 to to PART bracis-28398 255 6 allow allow VERB bracis-28398 255 7 attributing attribute VERB bracis-28398 255 8 a a DET bracis-28398 255 9 expected expect VERB bracis-28398 255 10 hardness hardness ADJ bracis-28398 255 11 level level NOUN bracis-28398 255 12 to to ADP bracis-28398 255 13 unlabeled unlabeled ADJ bracis-28398 255 14 test test NOUN bracis-28398 255 15 instances instance NOUN bracis-28398 255 16 . . PUNCT bracis-28398 256 1 also also ADV bracis-28398 256 2 , , PUNCT bracis-28398 256 3 applying apply VERB bracis-28398 256 4 the the DET bracis-28398 256 5 analysis analysis NOUN bracis-28398 256 6 to to ADP bracis-28398 256 7 imbalanced imbalanced ADJ bracis-28398 256 8 datasets dataset NOUN bracis-28398 256 9 must must AUX bracis-28398 256 10 be be AUX bracis-28398 256 11 done do VERB bracis-28398 256 12 with with ADP bracis-28398 256 13 care care NOUN bracis-28398 256 14 , , PUNCT bracis-28398 256 15 since since SCONJ bracis-28398 256 16 the the DET bracis-28398 256 17 minority minority NOUN bracis-28398 256 18 class class NOUN bracis-28398 256 19 observations observation NOUN bracis-28398 256 20 will will AUX bracis-28398 256 21 tend tend VERB bracis-28398 256 22 to to PART bracis-28398 256 23 be be AUX bracis-28398 256 24 pointed point VERB bracis-28398 256 25 as as ADP bracis-28398 256 26 hard hard ADV bracis-28398 256 27 just just ADV bracis-28398 256 28 because because SCONJ bracis-28398 256 29 they they PRON bracis-28398 256 30 are be AUX bracis-28398 256 31 outnumbered outnumber VERB bracis-28398 256 32 . . PUNCT bracis-28398 257 1 and and CCONJ bracis-28398 257 2 missing miss VERB bracis-28398 257 3 values value NOUN bracis-28398 257 4 must must AUX bracis-28398 257 5 also also ADV bracis-28398 257 6 be be AUX bracis-28398 257 7 dealt deal VERB bracis-28398 257 8 with with ADP bracis-28398 257 9 in in ADP bracis-28398 257 10 advance advance NOUN bracis-28398 257 11 . . PUNCT bracis-28398 258 1 both both DET bracis-28398 258 2 issues issue NOUN bracis-28398 258 3 can can AUX bracis-28398 258 4 be be AUX bracis-28398 258 5 solved solve VERB bracis-28398 258 6 in in ADP bracis-28398 258 7 the the DET bracis-28398 258 8 future future NOUN bracis-28398 258 9 by by ADP bracis-28398 258 10 including include VERB bracis-28398 258 11 data datum NOUN bracis-28398 258 12 re re NOUN bracis-28398 258 13 - - NOUN bracis-28398 258 14 sampling sample VERB bracis-28398 258 15 and and CCONJ bracis-28398 258 16 missing miss VERB bracis-28398 258 17 value value NOUN bracis-28398 258 18 imputation imputation NOUN bracis-28398 258 19 strategies strategy NOUN bracis-28398 258 20 inside inside ADP bracis-28398 258 21 the the DET bracis-28398 258 22 framework framework NOUN bracis-28398 258 23 . . PUNCT bracis-28398 259 1 other other ADJ bracis-28398 259 2 types type NOUN bracis-28398 259 3 of of ADP bracis-28398 259 4 analysis analysis NOUN bracis-28398 259 5 can can AUX bracis-28398 259 6 also also ADV bracis-28398 259 7 be be AUX bracis-28398 259 8 devised devise VERB bracis-28398 259 9 and and CCONJ bracis-28398 259 10 investigated investigate VERB bracis-28398 259 11 . . PUNCT bracis-28398 260 1 for for ADP bracis-28398 260 2 instance instance NOUN bracis-28398 260 3 , , PUNCT bracis-28398 260 4 we we PRON bracis-28398 260 5 can can AUX bracis-28398 260 6 test test VERB bracis-28398 260 7 the the DET bracis-28398 260 8 effects effect NOUN bracis-28398 260 9 of of ADP bracis-28398 260 10 including include VERB bracis-28398 260 11 or or CCONJ bracis-28398 260 12 excluding exclude VERB bracis-28398 260 13 features feature NOUN bracis-28398 260 14 from from ADP bracis-28398 260 15 a a DET bracis-28398 260 16 dataset dataset NOUN bracis-28398 260 17 . . PUNCT bracis-28398 261 1 as as SCONJ bracis-28398 261 2 more more ADV bracis-28398 261 3 informative informative ADJ bracis-28398 261 4 features feature NOUN bracis-28398 261 5 are be AUX bracis-28398 261 6 present present ADJ bracis-28398 261 7 , , PUNCT bracis-28398 261 8 we we PRON bracis-28398 261 9 expect expect VERB bracis-28398 261 10 the the DET bracis-28398 261 11 easiness easiness ADJ bracis-28398 261 12 footprint footprint NOUN bracis-28398 261 13 to to PART bracis-28398 261 14 grow grow VERB bracis-28398 261 15 and and CCONJ bracis-28398 261 16 the the DET bracis-28398 261 17 hardness hardness ADJ bracis-28398 261 18 footprint footprint NOUN bracis-28398 261 19 to to PART bracis-28398 261 20 shrink shrink VERB bracis-28398 261 21 . . PUNCT bracis-28398 262 1 other other ADJ bracis-28398 262 2 analysis analysis NOUN bracis-28398 262 3 possible possible ADJ bracis-28398 262 4 is be AUX bracis-28398 262 5 to to PART bracis-28398 262 6 establish establish VERB bracis-28398 262 7 when when SCONJ bracis-28398 262 8 a a DET bracis-28398 262 9 prediction prediction NOUN bracis-28398 262 10 should should AUX bracis-28398 262 11 better well ADV bracis-28398 262 12 be be AUX bracis-28398 262 13 discarded discard VERB bracis-28398 262 14 and and CCONJ bracis-28398 262 15 be be AUX bracis-28398 262 16 entrusted entrust VERB bracis-28398 262 17 to to ADP bracis-28398 262 18 a a DET bracis-28398 262 19 specialist specialist NOUN bracis-28398 262 20 instead instead ADV bracis-28398 262 21 and and CCONJ bracis-28398 262 22 to to PART bracis-28398 262 23 devise devise VERB bracis-28398 262 24 a a DET bracis-28398 262 25 human human NOUN bracis-28398 262 26 - - PUNCT bracis-28398 262 27 in in ADP bracis-28398 262 28 - - PUNCT bracis-28398 262 29 the the DET bracis-28398 262 30 - - PUNCT bracis-28398 262 31 loop loop NOUN bracis-28398 262 32 procedure procedure NOUN bracis-28398 262 33 . . PUNCT bracis-28398 263 1 notes note VERB bracis-28398 263 2 1.(https://github.com 1.(https://github.com PROPN bracis-28398 263 3 / / SYM bracis-28398 263 4 gabivaleriano gabivaleriano PROPN bracis-28398 263 5 / / SYM bracis-28398 263 6 explaining_healthdata explaining_healthdata PROPN bracis-28398 263 7 ) ) PUNCT bracis-28398 263 8 . . PUNCT bracis-28398 264 1 references reference NOUN bracis-28398 264 2 anderson anderson PROPN bracis-28398 264 3 , , PUNCT bracis-28398 264 4 d. d. PROPN bracis-28398 264 5 , , PUNCT bracis-28398 264 6 bjarnadottir bjarnadottir VERB bracis-28398 264 7 , , PUNCT bracis-28398 264 8 m.v m.v PROPN bracis-28398 264 9 . . PROPN bracis-28398 264 10 , , PUNCT bracis-28398 264 11 nenova nenova PROPN bracis-28398 264 12 , , PUNCT bracis-28398 264 13 z. z. PROPN bracis-28398 264 14 : : PUNCT bracis-28398 264 15 machine machine NOUN bracis-28398 264 16 learning learning NOUN bracis-28398 264 17 in in ADP bracis-28398 264 18 healthcare healthcare NOUN bracis-28398 264 19 : : PUNCT bracis-28398 264 20 operational operational ADJ bracis-28398 264 21 and and CCONJ bracis-28398 264 22 financial financial ADJ bracis-28398 264 23 impact impact NOUN bracis-28398 264 24 . . PUNCT bracis-28398 265 1 in in ADP bracis-28398 265 2 : : PUNCT bracis-28398 265 3 babich babich PROPN bracis-28398 265 4 , , PUNCT bracis-28398 265 5 v. v. ADV bracis-28398 265 6 , , PUNCT bracis-28398 265 7 birge birge NOUN bracis-28398 265 8 , , PUNCT bracis-28398 265 9 j.r j.r PROPN bracis-28398 265 10 . . PROPN bracis-28398 265 11 , , PUNCT bracis-28398 265 12 hilary hilary PROPN bracis-28398 265 13 , , PUNCT bracis-28398 265 14 g. g. PROPN bracis-28398 265 15 ( ( PUNCT bracis-28398 265 16 eds eds PROPN bracis-28398 265 17 . . PUNCT bracis-28398 265 18 ) ) PUNCT bracis-28398 265 19 innovative innovative ADJ bracis-28398 265 20 technology technology NOUN bracis-28398 265 21 at at ADP bracis-28398 265 22 the the DET bracis-28398 265 23 interface interface NOUN bracis-28398 265 24 of of ADP bracis-28398 265 25 finance finance NOUN bracis-28398 265 26 and and CCONJ bracis-28398 265 27 operations operation NOUN bracis-28398 265 28 , , PUNCT bracis-28398 265 29 vol vol NOUN bracis-28398 265 30 . . PROPN bracis-28398 265 31 11 11 NUM bracis-28398 265 32 , , PUNCT bracis-28398 265 33 pp pp ADJ bracis-28398 265 34 . . PUNCT bracis-28398 266 1 153–174 153–174 NUM bracis-28398 266 2 . . PUNCT bracis-28398 266 3 springer springer NOUN bracis-28398 266 4 , , PUNCT bracis-28398 266 5 cham cham PROPN bracis-28398 266 6 ( ( PUNCT bracis-28398 266 7 2022 2022 NUM bracis-28398 266 8 ) ) PUNCT bracis-28398 266 9 . . PUNCT bracis-28398 267 1 https://doi.org/10.1007/978-3-030-75729-8_5 https://doi.org/10.1007/978-3-030-75729-8_5 PROPN bracis-28398 267 2 imrie imrie PROPN bracis-28398 267 3 , , PUNCT bracis-28398 267 4 f. f. PROPN bracis-28398 267 5 , , PUNCT bracis-28398 267 6 cebere cebere ADV bracis-28398 267 7 , , PUNCT bracis-28398 267 8 b. b. PROPN bracis-28398 267 9 , , PUNCT bracis-28398 267 10 mckinney mckinney PROPN bracis-28398 267 11 , , PUNCT bracis-28398 267 12 e.f e.f PROPN bracis-28398 267 13 . . PROPN bracis-28398 267 14 , , PUNCT bracis-28398 267 15 van van PROPN bracis-28398 267 16 der der NOUN bracis-28398 267 17 schaar schaar NOUN bracis-28398 267 18 m. m. NOUN bracis-28398 267 19 : : PUNCT bracis-28398 267 20 autoprognosis autoprognosis NOUN bracis-28398 267 21 2.0 2.0 NUM bracis-28398 267 22 : : PUNCT bracis-28398 267 23 democratizing democratize VERB bracis-28398 267 24 diagnostic diagnostic ADJ bracis-28398 267 25 and and CCONJ bracis-28398 267 26 prognostic prognostic ADJ bracis-28398 267 27 modeling modeling NOUN bracis-28398 267 28 in in ADP bracis-28398 267 29 healthcare healthcare NOUN bracis-28398 267 30 with with ADP bracis-28398 267 31 automated automate VERB bracis-28398 267 32 machine machine NOUN bracis-28398 267 33 learning learning NOUN bracis-28398 267 34 . . PUNCT bracis-28398 268 1 arxiv arxiv PROPN bracis-28398 268 2 preprint preprint NOUN bracis-28398 268 3 arxiv:2210.12090 arxiv:2210.12090 NOUN bracis-28398 268 4 ( ( PUNCT bracis-28398 268 5 2022 2022 NUM bracis-28398 268 6 ) ) PUNCT bracis-28398 268 7 de de X bracis-28398 268 8 moraes moraes PROPN bracis-28398 268 9 , , PUNCT bracis-28398 268 10 b.a.f b.a.f PROPN bracis-28398 268 11 . . PROPN bracis-28398 268 12 , , PUNCT bracis-28398 268 13 miraglia miraglia PROPN bracis-28398 268 14 , , PUNCT bracis-28398 268 15 j. j. PROPN bracis-28398 268 16 , , PUNCT bracis-28398 268 17 donato donato PROPN bracis-28398 268 18 , , PUNCT bracis-28398 268 19 t. t. PROPN bracis-28398 268 20 , , PUNCT bracis-28398 268 21 filho filho PROPN bracis-28398 268 22 , , PUNCT bracis-28398 268 23 a. a. NOUN bracis-28398 268 24 : : PUNCT bracis-28398 268 25 covid-19 covid-19 PROPN bracis-28398 268 26 diagnosis diagnosis NOUN bracis-28398 268 27 prediction prediction NOUN bracis-28398 268 28 in in ADP bracis-28398 268 29 emergency emergency NOUN bracis-28398 268 30 care care NOUN bracis-28398 268 31 patients patient NOUN bracis-28398 268 32 : : PUNCT bracis-28398 268 33 a a DET bracis-28398 268 34 machine machine NOUN bracis-28398 268 35 learning learn VERB bracis-28398 268 36 approach approach NOUN bracis-28398 268 37 . . PUNCT bracis-28398 269 1 medrxiv medrxiv PROPN bracis-28398 269 2 , , PUNCT bracis-28398 269 3 2020 2020 NUM bracis-28398 269 4 - - SYM bracis-28398 269 5 04 04 NUM bracis-28398 269 6 ( ( PUNCT bracis-28398 269 7 2020 2020 NUM bracis-28398 269 8 ) ) PUNCT bracis-28398 269 9 google google PROPN bracis-28398 269 10 scholar scholar NOUN bracis-28398 269 11     SPACE bracis-28398 269 12 fernandes fernande NOUN bracis-28398 269 13 , , PUNCT bracis-28398 269 14 f.t f.t PROPN bracis-28398 269 15 . . PROPN bracis-28398 269 16 , , PUNCT bracis-28398 269 17 de de PROPN bracis-28398 269 18 oliveira oliveira PROPN bracis-28398 269 19 , , PUNCT bracis-28398 269 20 t.a t.a PROPN bracis-28398 269 21 . . PROPN bracis-28398 269 22 , , PUNCT bracis-28398 269 23 teixeira teixeira PROPN bracis-28398 269 24 , , PUNCT bracis-28398 269 25 c.e c.e PROPN bracis-28398 269 26 . . PROPN bracis-28398 269 27 , , PUNCT bracis-28398 269 28 de de PROPN bracis-28398 269 29 moraes moraes PROPN bracis-28398 269 30 batista batista PROPN bracis-28398 269 31 , , PUNCT bracis-28398 269 32 a.f a.f PROPN bracis-28398 269 33 . . PROPN bracis-28398 269 34 , , PUNCT bracis-28398 269 35 dalla dalla PROPN bracis-28398 269 36 costa costa PROPN bracis-28398 269 37 , , PUNCT bracis-28398 269 38 g. g. PROPN bracis-28398 269 39 , , PUNCT bracis-28398 269 40 chiavegatto chiavegatto NOUN bracis-28398 269 41 filho filho NOUN bracis-28398 269 42 , , PUNCT bracis-28398 269 43 a.d.p a.d.p PROPN bracis-28398 269 44 . . PROPN bracis-28398 269 45 : : PUNCT bracis-28398 269 46 a a DET bracis-28398 269 47 multipurpose multipurpose NOUN bracis-28398 269 48 machine machine NOUN bracis-28398 269 49 learning learn VERB bracis-28398 269 50 approach approach NOUN bracis-28398 269 51 to to PART bracis-28398 269 52 predict predict VERB bracis-28398 269 53 covid-19 covid-19 PROPN bracis-28398 269 54 negative negative ADJ bracis-28398 269 55 prognosis prognosis NOUN bracis-28398 269 56 in in ADP bracis-28398 269 57 são são PROPN bracis-28398 269 58 paulo paulo PROPN bracis-28398 269 59 , , PUNCT bracis-28398 269 60 brazil brazil PROPN bracis-28398 269 61 . . PUNCT bracis-28398 270 1 sci sci PROPN bracis-28398 270 2 . . PROPN bracis-28398 270 3 rep rep PROPN bracis-28398 270 4 . . PROPN bracis-28398 270 5 11(1 11(1 NUM bracis-28398 270 6 ) ) PUNCT bracis-28398 270 7 , , PUNCT bracis-28398 270 8 1–7 1–7 NUM bracis-28398 270 9 ( ( PUNCT bracis-28398 270 10 2021 2021 NUM bracis-28398 270 11 ) ) PUNCT bracis-28398 270 12 article article NOUN bracis-28398 270 13     SPACE bracis-28398 270 14 google google PROPN bracis-28398 270 15 scholar scholar NOUN bracis-28398 270 16     SPACE bracis-28398 270 17 wynants wynant NOUN bracis-28398 270 18 , , PUNCT bracis-28398 270 19 l. l. PROPN bracis-28398 270 20 , , PUNCT bracis-28398 270 21 et et PROPN bracis-28398 270 22 al al PROPN bracis-28398 270 23 . . PROPN bracis-28398 270 24 : : PUNCT bracis-28398 271 1 prediction prediction NOUN bracis-28398 271 2 models model NOUN bracis-28398 271 3 for for ADP bracis-28398 271 4 diagnosis diagnosis NOUN bracis-28398 271 5 and and CCONJ bracis-28398 271 6 prognosis prognosis NOUN bracis-28398 271 7 of of ADP bracis-28398 271 8 covid-19 covid-19 PROPN bracis-28398 271 9 : : PUNCT bracis-28398 271 10 systematic systematic ADJ bracis-28398 271 11 review review NOUN bracis-28398 271 12 and and CCONJ bracis-28398 271 13 critical critical ADJ bracis-28398 271 14 appraisal appraisal NOUN bracis-28398 271 15 . . PUNCT bracis-28398 272 1 bmj bmj PROPN bracis-28398 272 2 369 369 NUM bracis-28398 272 3 ( ( PUNCT bracis-28398 272 4 2020 2020 NUM bracis-28398 272 5 ) ) PUNCT bracis-28398 272 6 . . PUNCT bracis-28398 273 1 https://doi.org/10.1136/bmj.m1328 https://doi.org/10.1136/bmj.m1328 NUM bracis-28398 273 2 seedat seedat NOUN bracis-28398 273 3 , , PUNCT bracis-28398 273 4 n. n. NOUN bracis-28398 273 5 , , PUNCT bracis-28398 273 6 crabbe crabbe PROPN bracis-28398 273 7 j. j. PROPN bracis-28398 273 8 , , PUNCT bracis-28398 273 9 van van PROPN bracis-28398 273 10 der der ADJ bracis-28398 273 11 schaar schaar NOUN bracis-28398 273 12 , , PUNCT bracis-28398 273 13 m. m. NOUN bracis-28398 273 14 : : PUNCT bracis-28398 273 15 data data NOUN bracis-28398 273 16 - - PUNCT bracis-28398 273 17 suite suite NOUN bracis-28398 273 18 : : PUNCT bracis-28398 273 19 data data NOUN bracis-28398 273 20 - - PUNCT bracis-28398 273 21 centric centric ADJ bracis-28398 273 22 identification identification NOUN bracis-28398 273 23 of of ADP bracis-28398 273 24 in in ADP bracis-28398 273 25 - - PUNCT bracis-28398 273 26 distribution distribution NOUN bracis-28398 273 27 incongruous incongruous ADJ bracis-28398 273 28 examples example NOUN bracis-28398 273 29 . . PUNCT bracis-28398 274 1 arxiv arxiv PROPN bracis-28398 274 2 preprint preprint NOUN bracis-28398 274 3 arxiv:2202.08836 arxiv:2202.08836 NOUN bracis-28398 274 4 ( ( PUNCT bracis-28398 274 5 2022 2022 NUM bracis-28398 274 6 ) ) PUNCT bracis-28398 274 7 seedat seedat NOUN bracis-28398 274 8 , , PUNCT bracis-28398 274 9 n. n. NOUN bracis-28398 274 10 , , PUNCT bracis-28398 274 11 crabbe crabbe PROPN bracis-28398 274 12 j. j. PROPN bracis-28398 274 13 , , PUNCT bracis-28398 274 14 bica bica PROPN bracis-28398 274 15 , , PUNCT bracis-28398 274 16 i. i. PROPN bracis-28398 274 17 , , PUNCT bracis-28398 274 18 van van PROPN bracis-28398 274 19 der der ADJ bracis-28398 274 20 schaar schaar NOUN bracis-28398 274 21 , , PUNCT bracis-28398 274 22 m. m. NOUN bracis-28398 274 23 : : PUNCT bracis-28398 274 24 data data PROPN bracis-28398 274 25 - - PUNCT bracis-28398 274 26 iq iq PROPN bracis-28398 274 27 : : PUNCT bracis-28398 274 28 characterizing characterize VERB bracis-28398 274 29 subgroups subgroup NOUN bracis-28398 274 30 with with ADP bracis-28398 274 31 heterogeneous heterogeneous ADJ bracis-28398 274 32 outcomes outcome NOUN bracis-28398 274 33 in in ADP bracis-28398 274 34 tabular tabular PROPN bracis-28398 274 35 data data PROPN bracis-28398 274 36 . . PUNCT bracis-28398 275 1 arxiv arxiv PROPN bracis-28398 275 2 preprint preprint NOUN bracis-28398 275 3 arxiv:2210.13043 arxiv:2210.13043 NOUN bracis-28398 275 4 ( ( PUNCT bracis-28398 275 5 2022 2022 NUM bracis-28398 275 6 ) ) PUNCT bracis-28398 275 7 paiva paiva NOUN bracis-28398 275 8 , , PUNCT bracis-28398 275 9 p.y.a p.y.a NOUN bracis-28398 275 10 . . PROPN bracis-28398 275 11 , , PUNCT bracis-28398 275 12 moreno moreno PROPN bracis-28398 275 13 , , PUNCT bracis-28398 275 14 c.c c.c PROPN bracis-28398 275 15 . . PROPN bracis-28398 275 16 , , PUNCT bracis-28398 275 17 smith smith PROPN bracis-28398 275 18 - - PUNCT bracis-28398 275 19 miles mile NOUN bracis-28398 275 20 , , PUNCT bracis-28398 275 21 k. k. PROPN bracis-28398 275 22 , , PUNCT bracis-28398 275 23 valeriano valeriano PROPN bracis-28398 275 24 , , PUNCT bracis-28398 275 25 m.g m.g AUX bracis-28398 275 26 . . PROPN bracis-28398 275 27 , , PUNCT bracis-28398 275 28 lorena lorena PROPN bracis-28398 275 29 , , PUNCT bracis-28398 275 30 a.c a.c PROPN bracis-28398 275 31 . . PROPN bracis-28398 275 32 : : PUNCT bracis-28398 275 33 relating relate VERB bracis-28398 275 34 instance instance NOUN bracis-28398 275 35 hardness hardness NOUN bracis-28398 275 36 to to ADP bracis-28398 275 37 classification classification NOUN bracis-28398 275 38 performance performance NOUN bracis-28398 275 39 in in ADP bracis-28398 275 40 a a DET bracis-28398 275 41 dataset dataset NOUN bracis-28398 275 42 : : PUNCT bracis-28398 275 43 a a DET bracis-28398 275 44 visual visual ADJ bracis-28398 275 45 approach approach NOUN bracis-28398 275 46 . . PUNCT bracis-28398 276 1 mach mach NOUN bracis-28398 276 2 . . PUNCT bracis-28398 277 1 learn learn VERB bracis-28398 277 2 . . PUNCT bracis-28398 277 3 , , PUNCT bracis-28398 277 4 1–39 1–39 PROPN bracis-28398 277 5 ( ( PUNCT bracis-28398 277 6 2022 2022 NUM bracis-28398 277 7 ) ) PUNCT bracis-28398 277 8 google google PROPN bracis-28398 277 9 scholar scholar NOUN bracis-28398 277 10     SPACE bracis-28398 277 11 smith smith PROPN bracis-28398 277 12 , , PUNCT bracis-28398 277 13 m.r m.r PROPN bracis-28398 277 14 . . PROPN bracis-28398 277 15 , , PUNCT bracis-28398 277 16 martinez martinez PROPN bracis-28398 277 17 , , PUNCT bracis-28398 277 18 t. t. PROPN bracis-28398 277 19 , , PUNCT bracis-28398 277 20 giraud giraud NOUN bracis-28398 277 21 - - PUNCT bracis-28398 277 22 carrier carrier NOUN bracis-28398 277 23 , , PUNCT bracis-28398 277 24 c. c. PROPN bracis-28398 277 25 : : PUNCT bracis-28398 277 26 an an DET bracis-28398 277 27 instance instance NOUN bracis-28398 277 28 level level NOUN bracis-28398 277 29 analysis analysis NOUN bracis-28398 277 30 of of ADP bracis-28398 277 31 data datum NOUN bracis-28398 277 32 complexity complexity NOUN bracis-28398 277 33 . . PUNCT bracis-28398 278 1 mach mach NOUN bracis-28398 278 2 . . PUNCT bracis-28398 279 1 learn learn VERB bracis-28398 279 2 . . PUNCT bracis-28398 280 1 95(2 95(2 NOUN bracis-28398 280 2 ) ) PUNCT bracis-28398 280 3 , , PUNCT bracis-28398 280 4 225–256 225–256 NUM bracis-28398 280 5 ( ( PUNCT bracis-28398 280 6 2014 2014 NUM bracis-28398 280 7 ) ) PUNCT bracis-28398 280 8 article article NOUN bracis-28398 280 9     SPACE bracis-28398 280 10 mathscinet mathscinet NOUN bracis-28398 280 11     SPACE bracis-28398 280 12 math math NOUN bracis-28398 280 13     SPACE bracis-28398 280 14 google google PROPN bracis-28398 280 15 scholar scholar NOUN bracis-28398 280 16     SPACE bracis-28398 280 17 arruda arruda NOUN bracis-28398 280 18 , , PUNCT bracis-28398 280 19 j.l.m j.l.m PROPN bracis-28398 280 20 . . PROPN bracis-28398 280 21 , , PUNCT bracis-28398 280 22 prudêncio prudêncio PROPN bracis-28398 280 23 , , PUNCT bracis-28398 280 24 r.b.c r.b.c PROPN bracis-28398 280 25 . . PROPN bracis-28398 280 26 , , PUNCT bracis-28398 280 27 lorena lorena PROPN bracis-28398 280 28 , , PUNCT bracis-28398 280 29 a.c a.c PROPN bracis-28398 280 30 . . PROPN bracis-28398 280 31 : : PUNCT bracis-28398 281 1 measuring measure VERB bracis-28398 281 2 instance instance NOUN bracis-28398 281 3 hardness hardness NOUN bracis-28398 281 4 using use VERB bracis-28398 281 5 data datum NOUN bracis-28398 281 6 complexity complexity NOUN bracis-28398 281 7 measures measure NOUN bracis-28398 281 8 . . PUNCT bracis-28398 282 1 in in ADP bracis-28398 282 2 : : PUNCT bracis-28398 282 3 cerri cerri PROPN bracis-28398 282 4 , , PUNCT bracis-28398 282 5 r. r. PROPN bracis-28398 282 6 , , PUNCT bracis-28398 282 7 prati prati PROPN bracis-28398 282 8 , , PUNCT bracis-28398 282 9 r.c r.c PROPN bracis-28398 282 10 . . PROPN bracis-28398 282 11 ( ( PUNCT bracis-28398 282 12 eds ed NOUN bracis-28398 282 13 . . PUNCT bracis-28398 282 14 ) ) PUNCT bracis-28398 282 15 bracis bracis PROPN bracis-28398 282 16 2020 2020 NUM bracis-28398 282 17 . . PUNCT bracis-28398 283 1 lncs lncs PROPN bracis-28398 283 2 ( ( PUNCT bracis-28398 283 3 lnai lnai ADJ bracis-28398 283 4 ) ) PUNCT bracis-28398 283 5 , , PUNCT bracis-28398 283 6 vol vol NOUN bracis-28398 283 7 . . PROPN bracis-28398 283 8 12320 12320 NUM bracis-28398 283 9 , , PUNCT bracis-28398 283 10 pp pp ADJ bracis-28398 283 11 . . PUNCT bracis-28398 284 1 483–497 483–497 NUM bracis-28398 284 2 . . PUNCT bracis-28398 284 3 springer springer NOUN bracis-28398 284 4 , , PUNCT bracis-28398 284 5 cham cham PROPN bracis-28398 284 6 ( ( PUNCT bracis-28398 284 7 2020 2020 NUM bracis-28398 284 8 ) ) PUNCT bracis-28398 284 9 . . PUNCT bracis-28398 285 1 https://doi.org/10.1007/978-3-030-61380-8_33 https://doi.org/10.1007/978-3-030-61380-8_33 VERB bracis-28398 285 2 chapter chapter NOUN bracis-28398 285 3     SPACE bracis-28398 285 4 google google PROPN bracis-28398 285 5 scholar scholar NOUN bracis-28398 285 6     SPACE bracis-28398 285 7 paiva paiva NOUN bracis-28398 285 8 , , PUNCT bracis-28398 285 9 p.y.a p.y.a NOUN bracis-28398 285 10 . . PUNCT bracis-28398 285 11 , , PUNCT bracis-28398 285 12 smith smith PROPN bracis-28398 285 13 - - PUNCT bracis-28398 285 14 miles mile NOUN bracis-28398 285 15 , , PUNCT bracis-28398 285 16 k. k. PROPN bracis-28398 285 17 , , PUNCT bracis-28398 285 18 valeriano valeriano PROPN bracis-28398 285 19 , , PUNCT bracis-28398 285 20 m.g m.g PROPN bracis-28398 285 21 . . PROPN bracis-28398 285 22 , , PUNCT bracis-28398 285 23 lorena lorena PROPN bracis-28398 285 24 , , PUNCT bracis-28398 285 25 a.c a.c PROPN bracis-28398 285 26 . . PROPN bracis-28398 285 27 : : PUNCT bracis-28398 286 1 pyhard pyhard PROPN bracis-28398 286 2 : : PUNCT bracis-28398 286 3 a a DET bracis-28398 286 4 novel novel ADJ bracis-28398 286 5 tool tool NOUN bracis-28398 286 6 for for ADP bracis-28398 286 7 generating generate VERB bracis-28398 286 8 hardness hardness ADJ bracis-28398 286 9 embeddings embedding NOUN bracis-28398 286 10 to to PART bracis-28398 286 11 support support VERB bracis-28398 286 12 data datum NOUN bracis-28398 286 13 - - PUNCT bracis-28398 286 14 centric centric ADJ bracis-28398 286 15 analysis analysis NOUN bracis-28398 286 16 . . PUNCT bracis-28398 287 1 arxiv arxiv PROPN bracis-28398 287 2 preprint preprint NOUN bracis-28398 287 3 arxiv:2109.14430 arxiv:2109.14430 PROPN bracis-28398 287 4 ( ( PUNCT bracis-28398 287 5 2021 2021 NUM bracis-28398 287 6 ) ) PUNCT bracis-28398 287 7 valeriano valeriano PROPN bracis-28398 287 8 , , PUNCT bracis-28398 287 9 m.g m.g PROPN bracis-28398 287 10 . . PROPN bracis-28398 287 11 , , PUNCT bracis-28398 287 12 et et PROPN bracis-28398 287 13 al al PROPN bracis-28398 287 14 . . PUNCT bracis-28398 287 15 : : PUNCT bracis-28398 288 1 let let VERB bracis-28398 288 2 the the DET bracis-28398 288 3 data datum NOUN bracis-28398 288 4 speak speak VERB bracis-28398 288 5 : : PUNCT bracis-28398 288 6 analysing analyse VERB bracis-28398 288 7 data datum NOUN bracis-28398 288 8 from from ADP bracis-28398 288 9 multiple multiple ADJ bracis-28398 288 10 health health NOUN bracis-28398 288 11 centers center NOUN bracis-28398 288 12 of of ADP bracis-28398 288 13 the the DET bracis-28398 288 14 são são PROPN bracis-28398 288 15 paulo paulo PROPN bracis-28398 288 16 metropolitan metropolitan PROPN bracis-28398 288 17 area area NOUN bracis-28398 288 18 for for ADP bracis-28398 288 19 covid-19 covid-19 PROPN bracis-28398 288 20 clinical clinical ADJ bracis-28398 288 21 deterioration deterioration NOUN bracis-28398 288 22 prediction prediction NOUN bracis-28398 288 23 . . PUNCT bracis-28398 289 1 in in ADP bracis-28398 289 2 : : PUNCT bracis-28398 289 3 2022 2022 NUM bracis-28398 289 4 22nd 22nd NOUN bracis-28398 289 5 ieee ieee PROPN bracis-28398 289 6 international international ADJ bracis-28398 289 7 symposium symposium NOUN bracis-28398 289 8 on on ADP bracis-28398 289 9 cluster cluster NOUN bracis-28398 289 10 , , PUNCT bracis-28398 289 11 cloud cloud NOUN bracis-28398 289 12 and and CCONJ bracis-28398 289 13 internet internet NOUN bracis-28398 289 14 computing computing NOUN bracis-28398 289 15 ( ( PUNCT bracis-28398 289 16 ccgrid ccgrid NOUN bracis-28398 289 17 ) ) PUNCT bracis-28398 289 18 , , PUNCT bracis-28398 289 19 pp pp ADP bracis-28398 289 20 . . PUNCT bracis-28398 290 1 948–951 948–951 NUM bracis-28398 290 2 . . PUNCT bracis-28398 290 3 ieee ieee NOUN bracis-28398 290 4 ( ( PUNCT bracis-28398 290 5 2022 2022 NUM bracis-28398 290 6 ) ) PUNCT bracis-28398 290 7 google google PROPN bracis-28398 290 8 scholar scholar NOUN bracis-28398 290 9     SPACE bracis-28398 290 10 zheng zheng PROPN bracis-28398 290 11 , , PUNCT bracis-28398 290 12 k. k. PROPN bracis-28398 290 13 , , PUNCT bracis-28398 290 14 chen chen PROPN bracis-28398 290 15 , , PUNCT bracis-28398 290 16 g. g. PROPN bracis-28398 290 17 , , PUNCT bracis-28398 290 18 herschel herschel PROPN bracis-28398 290 19 , , PUNCT bracis-28398 290 20 m. m. NOUN bracis-28398 290 21 , , PUNCT bracis-28398 290 22 ngiam ngiam PROPN bracis-28398 290 23 , , PUNCT bracis-28398 290 24 k.y k.y PROPN bracis-28398 290 25 . . PROPN bracis-28398 290 26 , , PUNCT bracis-28398 290 27 ooi ooi PROPN bracis-28398 290 28 , , PUNCT bracis-28398 290 29 b.c b.c PROPN bracis-28398 290 30 . . PROPN bracis-28398 290 31 , , PUNCT bracis-28398 290 32 gao gao PROPN bracis-28398 290 33 , , PUNCT bracis-28398 290 34 j. j. PROPN bracis-28398 290 35 : : PUNCT bracis-28398 290 36 pace pace NOUN bracis-28398 290 37 : : PUNCT bracis-28398 290 38 learning learn VERB bracis-28398 290 39 effective effective ADJ bracis-28398 290 40 task task NOUN bracis-28398 290 41 decomposition decomposition NOUN bracis-28398 290 42 for for ADP bracis-28398 290 43 human human NOUN bracis-28398 290 44 - - PUNCT bracis-28398 290 45 in in ADP bracis-28398 290 46 - - PUNCT bracis-28398 290 47 the the DET bracis-28398 290 48 - - PUNCT bracis-28398 290 49 loop loop NOUN bracis-28398 290 50 healthcare healthcare PROPN bracis-28398 290 51 delivery delivery NOUN bracis-28398 290 52 . . PUNCT bracis-28398 291 1 in in ADP bracis-28398 291 2 : : PUNCT bracis-28398 291 3 proceedings proceeding NOUN bracis-28398 291 4 of of ADP bracis-28398 291 5 the the DET bracis-28398 291 6 2021 2021 NUM bracis-28398 291 7 international international ADJ bracis-28398 291 8 conference conference NOUN bracis-28398 291 9 on on ADP bracis-28398 291 10 management management NOUN bracis-28398 291 11 of of ADP bracis-28398 291 12 data datum NOUN bracis-28398 291 13 , , PUNCT bracis-28398 291 14 pp pp ADP bracis-28398 291 15 . . PUNCT bracis-28398 292 1 2156–2168 2156–2168 NUM bracis-28398 292 2 ( ( PUNCT bracis-28398 292 3 2021 2021 NUM bracis-28398 292 4 ) ) PUNCT bracis-28398 292 5 google google PROPN bracis-28398 292 6 scholar scholar NOUN bracis-28398 292 7     SPACE bracis-28398 292 8 houston houston PROPN bracis-28398 292 9 , , PUNCT bracis-28398 292 10 a. a. PROPN bracis-28398 292 11 , , PUNCT bracis-28398 292 12 cosma cosma PROPN bracis-28398 292 13 , , PUNCT bracis-28398 292 14 g. g. PROPN bracis-28398 292 15 , , PUNCT bracis-28398 292 16 turner turner PROPN bracis-28398 292 17 , , PUNCT bracis-28398 292 18 p. p. PROPN bracis-28398 292 19 , , PUNCT bracis-28398 292 20 bennett bennett PROPN bracis-28398 292 21 , , PUNCT bracis-28398 292 22 a. a. NOUN bracis-28398 292 23 : : PUNCT bracis-28398 292 24 predicting predict VERB bracis-28398 292 25 surgical surgical ADJ bracis-28398 292 26 outcomes outcome NOUN bracis-28398 292 27 for for ADP bracis-28398 292 28 chronic chronic ADJ bracis-28398 292 29 exertional exertional ADJ bracis-28398 292 30 compartment compartment NOUN bracis-28398 292 31 syndrome syndrome NOUN bracis-28398 292 32 using use VERB bracis-28398 292 33 a a DET bracis-28398 292 34 machine machine NOUN bracis-28398 292 35 learning learn VERB bracis-28398 292 36 framework framework NOUN bracis-28398 292 37 with with ADP bracis-28398 292 38 embedded embed VERB bracis-28398 292 39 trust trust NOUN bracis-28398 292 40 by by ADP bracis-28398 292 41 interrogation interrogation NOUN bracis-28398 292 42 strategies strategy NOUN bracis-28398 292 43 . . PUNCT bracis-28398 293 1 sci sci PROPN bracis-28398 293 2 . . PROPN bracis-28398 293 3 rep rep AUX bracis-28398 293 4 . . PROPN bracis-28398 293 5 11(1 11(1 NUM bracis-28398 293 6 ) ) PUNCT bracis-28398 293 7 , , PUNCT bracis-28398 293 8 1–15 1–15 PROPN bracis-28398 293 9 ( ( PUNCT bracis-28398 293 10 2021 2021 NUM bracis-28398 293 11 ) ) PUNCT bracis-28398 293 12 article article NOUN bracis-28398 293 13     SPACE bracis-28398 293 14 google google PROPN bracis-28398 293 15 scholar scholar NOUN bracis-28398 293 16     SPACE bracis-28398 293 17 prudêncio prudêncio NOUN bracis-28398 293 18 , , PUNCT bracis-28398 293 19 r.b r.b PROPN bracis-28398 293 20 . . PROPN bracis-28398 293 21 , , PUNCT bracis-28398 293 22 silva silva PROPN bracis-28398 293 23 filho filho PROPN bracis-28398 293 24 , , PUNCT bracis-28398 293 25 t.m t.m PROPN bracis-28398 293 26 . . PUNCT bracis-28398 293 27 : : PUNCT bracis-28398 293 28 explaining explain VERB bracis-28398 293 29 learning learn VERB bracis-28398 293 30 performance performance NOUN bracis-28398 293 31 with with ADP bracis-28398 293 32 local local ADJ bracis-28398 293 33 performance performance NOUN bracis-28398 293 34 regions region NOUN bracis-28398 293 35 and and CCONJ bracis-28398 293 36 maximally maximally ADV bracis-28398 293 37 relevant relevant ADJ bracis-28398 293 38 meta meta NOUN bracis-28398 293 39 - - PUNCT bracis-28398 293 40 rules rule NOUN bracis-28398 293 41 . . PUNCT bracis-28398 294 1 in in ADP bracis-28398 294 2 : : PUNCT bracis-28398 294 3 xavier xavier NOUN bracis-28398 294 4 - - PUNCT bracis-28398 294 5 junior junior PROPN bracis-28398 294 6 , , PUNCT bracis-28398 294 7 j.c j.c PROPN bracis-28398 294 8 . . PROPN bracis-28398 294 9 , , PUNCT bracis-28398 294 10 rios rios PROPN bracis-28398 294 11 , , PUNCT bracis-28398 294 12 r.a r.a PROPN bracis-28398 294 13 . . PROPN bracis-28398 294 14 ( ( PUNCT bracis-28398 294 15 eds ed NOUN bracis-28398 294 16 . . PUNCT bracis-28398 294 17 ) ) PUNCT bracis-28398 295 1 brazilian brazilian ADJ bracis-28398 295 2 conference conference NOUN bracis-28398 295 3 on on ADP bracis-28398 295 4 intelligent intelligent ADJ bracis-28398 295 5 systems system NOUN bracis-28398 295 6 , , PUNCT bracis-28398 295 7 pp pp ADJ bracis-28398 295 8 . . PUNCT bracis-28398 296 1 550–564 550–564 NUM bracis-28398 296 2 . . PUNCT bracis-28398 297 1 springer springer NOUN bracis-28398 297 2 , , PUNCT bracis-28398 297 3 cham cham PROPN bracis-28398 297 4 ( ( PUNCT bracis-28398 297 5 2022 2022 NUM bracis-28398 297 6 ) ) PUNCT bracis-28398 297 7 . . PUNCT bracis-28398 298 1 https://doi.org/10.1007/978-3-031-21686-2_38 https://doi.org/10.1007/978-3-031-21686-2_38 PROPN bracis-28398 298 2 gunning gunning NOUN bracis-28398 298 3 , , PUNCT bracis-28398 298 4 d. d. PROPN bracis-28398 298 5 , , PUNCT bracis-28398 298 6 stefik stefik VERB bracis-28398 298 7 , , PUNCT bracis-28398 298 8 m. m. NOUN bracis-28398 298 9 , , PUNCT bracis-28398 298 10 choi choi NOUN bracis-28398 298 11 , , PUNCT bracis-28398 298 12 j. j. PROPN bracis-28398 298 13 , , PUNCT bracis-28398 298 14 miller miller PROPN bracis-28398 298 15 , , PUNCT bracis-28398 298 16 t. t. PROPN bracis-28398 298 17 , , PUNCT bracis-28398 298 18 stumpf stumpf PROPN bracis-28398 298 19 , , PUNCT bracis-28398 298 20 s. s. PROPN bracis-28398 298 21 , , PUNCT bracis-28398 298 22 yang yang PROPN bracis-28398 298 23 , , PUNCT bracis-28398 298 24 g.-z g.-z PROPN bracis-28398 298 25 . . PUNCT bracis-28398 299 1 : : PUNCT bracis-28398 299 2 xai xai PROPN bracis-28398 299 3 - - PUNCT bracis-28398 299 4 explainable explainable ADJ bracis-28398 299 5 artificial artificial ADJ bracis-28398 299 6 intelligence intelligence NOUN bracis-28398 299 7 . . PUNCT bracis-28398 300 1 sci sci PROPN bracis-28398 300 2 . . PROPN bracis-28398 300 3 rob rob PROPN bracis-28398 300 4 . . PROPN bracis-28398 300 5 4(37 4(37 NUM bracis-28398 300 6 ) ) PUNCT bracis-28398 300 7 , , PUNCT bracis-28398 300 8 eaay7120 eaay7120 PROPN bracis-28398 300 9 ( ( PUNCT bracis-28398 300 10 2019 2019 NUM bracis-28398 300 11 ) ) PUNCT bracis-28398 300 12 google google NOUN bracis-28398 300 13 scholar scholar NOUN bracis-28398 300 14     SPACE bracis-28398 300 15 ojala ojala NOUN bracis-28398 300 16 , , PUNCT bracis-28398 300 17 m. m. NOUN bracis-28398 300 18 , , PUNCT bracis-28398 300 19 garriga garriga PROPN bracis-28398 300 20 , , PUNCT bracis-28398 300 21 g.c g.c PROPN bracis-28398 300 22 . . PROPN bracis-28398 300 23 : : PUNCT bracis-28398 300 24 permutation permutation NOUN bracis-28398 300 25 tests test VERB bracis-28398 300 26 for for ADP bracis-28398 300 27 studying study VERB bracis-28398 300 28 classifier classifier NOUN bracis-28398 300 29 performance performance NOUN bracis-28398 300 30 . . PUNCT bracis-28398 301 1 j. j. PROPN bracis-28398 301 2 mach mach PROPN bracis-28398 301 3 . . PUNCT bracis-28398 302 1 learn learn VERB bracis-28398 302 2 . . PUNCT bracis-28398 303 1 res re NOUN bracis-28398 303 2 . . PUNCT bracis-28398 304 1 11(6 11(6 NUM bracis-28398 304 2 ) ) PUNCT bracis-28398 304 3 ( ( PUNCT bracis-28398 304 4 2010 2010 NUM bracis-28398 304 5 ) ) PUNCT bracis-28398 304 6 google google PROPN bracis-28398 304 7 scholar scholar NOUN bracis-28398 304 8     SPACE bracis-28398 304 9 ghorbani ghorbani NOUN bracis-28398 304 10 , , PUNCT bracis-28398 304 11 a. a. NOUN bracis-28398 304 12 , , PUNCT bracis-28398 304 13 zou zou PROPN bracis-28398 304 14 , , PUNCT bracis-28398 304 15 j. j. PROPN bracis-28398 304 16 : : PUNCT bracis-28398 304 17 data data PROPN bracis-28398 304 18 shapley shapley NOUN bracis-28398 304 19 : : PUNCT bracis-28398 304 20 equitable equitable ADJ bracis-28398 304 21 valuation valuation NOUN bracis-28398 304 22 of of ADP bracis-28398 304 23 data datum NOUN bracis-28398 304 24 for for ADP bracis-28398 304 25 machine machine NOUN bracis-28398 304 26 learning learning NOUN bracis-28398 304 27 . . PUNCT bracis-28398 305 1 in in ADP bracis-28398 305 2 : : PUNCT bracis-28398 305 3 international international ADJ bracis-28398 305 4 conference conference NOUN bracis-28398 305 5 on on ADP bracis-28398 305 6 machine machine NOUN bracis-28398 305 7 learning learning NOUN bracis-28398 305 8 , , PUNCT bracis-28398 305 9 pp pp ADJ bracis-28398 305 10 . . PUNCT bracis-28398 305 11 2242–2251 2242–2251 NUM bracis-28398 305 12 . . PUNCT bracis-28398 305 13 pmlr pmlr NOUN bracis-28398 305 14 ( ( PUNCT bracis-28398 305 15 2019 2019 NUM bracis-28398 305 16 ) ) PUNCT bracis-28398 305 17 google google PROPN bracis-28398 305 18 scholar scholar NOUN bracis-28398 305 19     SPACE bracis-28398 305 20 lorena lorena PROPN bracis-28398 305 21 , , PUNCT bracis-28398 305 22 a.c a.c PROPN bracis-28398 305 23 . . PROPN bracis-28398 305 24 , , PUNCT bracis-28398 305 25 garcia garcia PROPN bracis-28398 305 26 , , PUNCT bracis-28398 305 27 l.p l.p PROPN bracis-28398 305 28 . . PROPN bracis-28398 305 29 , , PUNCT bracis-28398 305 30 lehmann lehmann PROPN bracis-28398 305 31 , , PUNCT bracis-28398 305 32 j. j. PROPN bracis-28398 305 33 , , PUNCT bracis-28398 305 34 souto souto PROPN bracis-28398 305 35 , , PUNCT bracis-28398 305 36 m.c m.c PROPN bracis-28398 305 37 . . PROPN bracis-28398 305 38 , , PUNCT bracis-28398 305 39 ho ho PROPN bracis-28398 305 40 , , PUNCT bracis-28398 305 41 t.k t.k PROPN bracis-28398 305 42 . . PROPN bracis-28398 305 43 : : PUNCT bracis-28398 305 44 how how SCONJ bracis-28398 305 45 complex complex ADJ bracis-28398 305 46 is be AUX bracis-28398 305 47 your your PRON bracis-28398 305 48 classification classification NOUN bracis-28398 305 49 problem problem NOUN bracis-28398 305 50 ? ? PUNCT bracis-28398 306 1 a a DET bracis-28398 306 2 survey survey NOUN bracis-28398 306 3 on on ADP bracis-28398 306 4 measuring measure VERB bracis-28398 306 5 classification classification NOUN bracis-28398 306 6 complexity complexity NOUN bracis-28398 306 7 . . PUNCT bracis-28398 307 1 acm acm PROPN bracis-28398 307 2 comput comput NOUN bracis-28398 307 3 . . PUNCT bracis-28398 308 1 surv surv NOUN bracis-28398 308 2 . . PUNCT bracis-28398 309 1 52(5 52(5 NUM bracis-28398 309 2 ) ) PUNCT bracis-28398 309 3 , , PUNCT bracis-28398 309 4 1–34 1–34 NUM bracis-28398 309 5 ( ( PUNCT bracis-28398 309 6 2019 2019 NUM bracis-28398 309 7 ) ) PUNCT bracis-28398 309 8 article article NOUN bracis-28398 309 9     SPACE bracis-28398 309 10 google google PROPN bracis-28398 309 11 scholar scholar NOUN bracis-28398 309 12     SPACE bracis-28398 309 13 jafarzadeh jafarzadeh ADJ bracis-28398 309 14 , , PUNCT bracis-28398 309 15 a. a. NOUN bracis-28398 309 16 , , PUNCT bracis-28398 309 17 jafarzadeh jafarzadeh PROPN bracis-28398 309 18 , , PUNCT bracis-28398 309 19 s. s. PROPN bracis-28398 309 20 , , PUNCT bracis-28398 309 21 nozari nozari PROPN bracis-28398 309 22 , , PUNCT bracis-28398 309 23 p. p. NOUN bracis-28398 309 24 , , PUNCT bracis-28398 309 25 mokhtari mokhtari PROPN bracis-28398 309 26 , , PUNCT bracis-28398 309 27 p. p. NOUN bracis-28398 309 28 , , PUNCT bracis-28398 309 29 nemati nemati PROPN bracis-28398 309 30 , , PUNCT bracis-28398 309 31 m. m. NOUN bracis-28398 309 32 : : PUNCT bracis-28398 309 33 lymphopenia lymphopenia NOUN bracis-28398 309 34 an an DET bracis-28398 309 35 important important ADJ bracis-28398 309 36 immunological immunological ADJ bracis-28398 309 37 abnormality abnormality NOUN bracis-28398 309 38 in in ADP bracis-28398 309 39 patients patient NOUN bracis-28398 309 40 with with ADP bracis-28398 309 41 covid-19 covid-19 PROPN bracis-28398 309 42 : : PUNCT bracis-28398 309 43 possible possible ADJ bracis-28398 309 44 mechanisms mechanism NOUN bracis-28398 309 45 . . PUNCT bracis-28398 310 1 scand scand PROPN bracis-28398 310 2 . . PUNCT bracis-28398 311 1 j. j. PROPN bracis-28398 311 2 immunol immunol PROPN bracis-28398 311 3 . . PUNCT bracis-28398 312 1 93(2 93(2 NOUN bracis-28398 312 2 ) ) PUNCT bracis-28398 312 3 , , PUNCT bracis-28398 312 4 e12967 e12967 NOUN bracis-28398 312 5 ( ( PUNCT bracis-28398 312 6 2021 2021 NUM bracis-28398 312 7 ) ) PUNCT bracis-28398 312 8 article article NOUN bracis-28398 312 9     SPACE bracis-28398 312 10 google google PROPN bracis-28398 312 11 scholar scholar NOUN bracis-28398 312 12     SPACE bracis-28398 312 13 amankwaa amankwaa NOUN bracis-28398 312 14 - - PUNCT bracis-28398 312 15 kyeremeh kyeremeh NOUN bracis-28398 312 16 , , PUNCT bracis-28398 312 17 b. b. PROPN bracis-28398 312 18 , , PUNCT bracis-28398 312 19 greet greet VERB bracis-28398 312 20 , , PUNCT bracis-28398 312 21 c. c. PROPN bracis-28398 312 22 , , PUNCT bracis-28398 312 23 zanin zanin PROPN bracis-28398 312 24 , , PUNCT bracis-28398 312 25 m. m. NOUN bracis-28398 312 26 , , PUNCT bracis-28398 312 27 skinner skinner PROPN bracis-28398 312 28 , , PUNCT bracis-28398 312 29 w. w. PROPN bracis-28398 312 30 , , PUNCT bracis-28398 312 31 asamoah asamoah PROPN bracis-28398 312 32 , , PUNCT bracis-28398 312 33 r.k r.k PROPN bracis-28398 312 34 . . PROPN bracis-28398 312 35 : : PUNCT bracis-28398 312 36 selecting select VERB bracis-28398 312 37 key key ADJ bracis-28398 312 38 predictor predictor NOUN bracis-28398 312 39 parameters parameter NOUN bracis-28398 312 40 for for ADP bracis-28398 312 41 regression regression NOUN bracis-28398 312 42 analysis analysis NOUN bracis-28398 312 43 using use VERB bracis-28398 312 44 modified modify VERB bracis-28398 312 45 neighbourhood neighbourhood NOUN bracis-28398 312 46 component component NOUN bracis-28398 312 47 analysis analysis NOUN bracis-28398 312 48 ( ( PUNCT bracis-28398 312 49 nca nca NOUN bracis-28398 312 50 ) ) PUNCT bracis-28398 312 51 algorithm algorithm NOUN bracis-28398 312 52 . . PUNCT bracis-28398 313 1 in in ADP bracis-28398 313 2 : : PUNCT bracis-28398 313 3 proceedings proceeding NOUN bracis-28398 313 4 of of ADP bracis-28398 313 5 6th 6th ADJ bracis-28398 313 6 umat umat ADJ bracis-28398 313 7 biennial biennial ADJ bracis-28398 313 8 international international ADJ bracis-28398 313 9 mining mining NOUN bracis-28398 313 10 and and CCONJ bracis-28398 313 11 mineral mineral NOUN bracis-28398 313 12 conference conference NOUN bracis-28398 313 13 , , PUNCT bracis-28398 313 14 pp pp X bracis-28398 313 15 . . PUNCT bracis-28398 314 1 320–325 320–325 NUM bracis-28398 314 2 ( ( PUNCT bracis-28398 314 3 2020 2020 NUM bracis-28398 314 4 ) ) PUNCT bracis-28398 314 5 google google PROPN bracis-28398 314 6 scholar scholar NOUN bracis-28398 314 7     SPACE bracis-28398 314 8 smith smith PROPN bracis-28398 314 9 - - PUNCT bracis-28398 314 10 miles mile NOUN bracis-28398 314 11 , , PUNCT bracis-28398 314 12 k. k. PROPN bracis-28398 314 13 , , PUNCT bracis-28398 314 14 tan tan PROPN bracis-28398 314 15 , , PUNCT bracis-28398 314 16 t.t t.t PROPN bracis-28398 314 17 . . PROPN bracis-28398 314 18 : : PUNCT bracis-28398 314 19 measuring measure VERB bracis-28398 314 20 algorithm algorithm NOUN bracis-28398 314 21 footprints footprint NOUN bracis-28398 314 22 in in ADP bracis-28398 314 23 instance instance NOUN bracis-28398 314 24 space space NOUN bracis-28398 314 25 . . PUNCT bracis-28398 315 1 in in ADP bracis-28398 315 2 : : PUNCT bracis-28398 315 3 2012 2012 NUM bracis-28398 315 4 ieee ieee PROPN bracis-28398 315 5 congress congress PROPN bracis-28398 315 6 on on ADP bracis-28398 315 7 evolutionary evolutionary ADJ bracis-28398 315 8 computation computation NOUN bracis-28398 315 9 , , PUNCT bracis-28398 315 10 pp pp ADP bracis-28398 315 11 . . PUNCT bracis-28398 315 12 1–8 1–8 X bracis-28398 315 13 . . PUNCT bracis-28398 315 14 ieee ieee PROPN bracis-28398 315 15 ( ( PUNCT bracis-28398 315 16 2012 2012 NUM bracis-28398 315 17 ) ) PUNCT bracis-28398 315 18 google google PROPN bracis-28398 315 19 scholar scholar NOUN bracis-28398 315 20     SPACE bracis-28398 315 21 muñoz muñoz PROPN bracis-28398 315 22 , , PUNCT bracis-28398 315 23 m.a m.a PROPN bracis-28398 315 24 . . PROPN bracis-28398 315 25 , , PUNCT bracis-28398 315 26 villanova villanova PROPN bracis-28398 315 27 , , PUNCT bracis-28398 315 28 l. l. PROPN bracis-28398 315 29 , , PUNCT bracis-28398 315 30 baatar baatar NOUN bracis-28398 315 31 , , PUNCT bracis-28398 315 32 d. d. PROPN bracis-28398 315 33 , , PUNCT bracis-28398 315 34 smith smith PROPN bracis-28398 315 35 - - PUNCT bracis-28398 315 36 miles mile NOUN bracis-28398 315 37 , , PUNCT bracis-28398 315 38 k. k. PROPN bracis-28398 315 39 : : PUNCT bracis-28398 315 40 instance instance NOUN bracis-28398 315 41 spaces space VERB bracis-28398 315 42 for for ADP bracis-28398 315 43 machine machine NOUN bracis-28398 315 44 learning learn VERB bracis-28398 315 45 classification classification NOUN bracis-28398 315 46 . . PUNCT bracis-28398 316 1 mach mach NOUN bracis-28398 316 2 . . PUNCT bracis-28398 317 1 learn learn VERB bracis-28398 317 2 . . PUNCT bracis-28398 318 1 107(1 107(1 NUM bracis-28398 318 2 ) ) PUNCT bracis-28398 318 3 , , PUNCT bracis-28398 318 4 109–147 109–147 NUM bracis-28398 318 5 ( ( PUNCT bracis-28398 318 6 2018 2018 NUM bracis-28398 318 7 ) ) PUNCT bracis-28398 318 8 article article NOUN bracis-28398 318 9     SPACE bracis-28398 318 10 mathscinet mathscinet NOUN bracis-28398 318 11     SPACE bracis-28398 318 12 math math NOUN bracis-28398 318 13     SPACE bracis-28398 318 14 google google PROPN bracis-28398 318 15 scholar scholar NOUN bracis-28398 318 16     SPACE bracis-28398 318 17 khan khan PROPN bracis-28398 318 18 , , PUNCT bracis-28398 318 19 k. k. PROPN bracis-28398 318 20 , , PUNCT bracis-28398 318 21 rehman rehman PROPN bracis-28398 318 22 , , PUNCT bracis-28398 318 23 s.u s.u PROPN bracis-28398 318 24 . . PROPN bracis-28398 318 25 , , PUNCT bracis-28398 318 26 aziz aziz PROPN bracis-28398 318 27 , , PUNCT bracis-28398 318 28 k. k. PROPN bracis-28398 318 29 , , PUNCT bracis-28398 318 30 fong fong PROPN bracis-28398 318 31 , , PUNCT bracis-28398 318 32 s. s. PROPN bracis-28398 318 33 , , PUNCT bracis-28398 318 34 sarasvady sarasvady PROPN bracis-28398 318 35 , , PUNCT bracis-28398 318 36 s. s. PROPN bracis-28398 318 37 : : PUNCT bracis-28398 318 38 dbscan dbscan NOUN bracis-28398 318 39 : : PUNCT bracis-28398 318 40 past past ADJ bracis-28398 318 41 , , PUNCT bracis-28398 318 42 present present ADJ bracis-28398 318 43 and and CCONJ bracis-28398 318 44 future future ADJ bracis-28398 318 45 . . PUNCT bracis-28398 319 1 in in ADP bracis-28398 319 2 : : PUNCT bracis-28398 319 3 the the DET bracis-28398 319 4 fifth fifth ADJ bracis-28398 319 5 international international ADJ bracis-28398 319 6 conference conference NOUN bracis-28398 319 7 on on ADP bracis-28398 319 8 the the DET bracis-28398 319 9 applications application NOUN bracis-28398 319 10 of of ADP bracis-28398 319 11 digital digital ADJ bracis-28398 319 12 information information NOUN bracis-28398 319 13 and and CCONJ bracis-28398 319 14 web web NOUN bracis-28398 319 15 technologies technology NOUN bracis-28398 319 16 , , PUNCT bracis-28398 319 17 pp pp ADV bracis-28398 319 18 . . PUNCT bracis-28398 320 1 232–238 232–238 NUM bracis-28398 320 2 . . X bracis-28398 320 3 ieee ieee NOUN bracis-28398 320 4 ( ( PUNCT bracis-28398 320 5 2014 2014 NUM bracis-28398 320 6 ) ) PUNCT bracis-28398 320 7 google google NOUN bracis-28398 320 8 scholar scholar NOUN bracis-28398 320 9     SPACE bracis-28398 320 10 edelsbrunner edelsbrunner NOUN bracis-28398 320 11 , , PUNCT bracis-28398 320 12 h. h. NOUN bracis-28398 320 13 : : PUNCT bracis-28398 320 14 alpha alpha NOUN bracis-28398 320 15 shapes shape NOUN bracis-28398 320 16 - - PUNCT bracis-28398 320 17 a a DET bracis-28398 320 18 survey survey NOUN bracis-28398 320 19 . . PUNCT bracis-28398 321 1 tessellations tessellation NOUN bracis-28398 321 2 sci sci PROPN bracis-28398 321 3 . . PROPN bracis-28398 322 1 27 27 NUM bracis-28398 322 2 , , PUNCT bracis-28398 322 3 1–25 1–25 PROPN bracis-28398 322 4 ( ( PUNCT bracis-28398 322 5 2010 2010 NUM bracis-28398 322 6 ) ) PUNCT bracis-28398 322 7 google google PROPN bracis-28398 322 8 scholar scholar NOUN bracis-28398 322 9     SPACE bracis-28398 322 10 najafabadi najafabadi NOUN bracis-28398 322 11 , , PUNCT bracis-28398 322 12 m.m m.m PROPN bracis-28398 322 13 . . PROPN bracis-28398 322 14 , , PUNCT bracis-28398 322 15 villanustre villanustre PROPN bracis-28398 322 16 , , PUNCT bracis-28398 322 17 f. f. PROPN bracis-28398 322 18 , , PUNCT bracis-28398 322 19 khoshgoftaar khoshgoftaar PROPN bracis-28398 322 20 , , PUNCT bracis-28398 322 21 t.m t.m PROPN bracis-28398 322 22 . . PROPN bracis-28398 322 23 , , PUNCT bracis-28398 322 24 seliya seliya PROPN bracis-28398 322 25 , , PUNCT bracis-28398 322 26 n. n. NOUN bracis-28398 322 27 , , PUNCT bracis-28398 322 28 wald wald PROPN bracis-28398 322 29 , , PUNCT bracis-28398 322 30 r. r. PROPN bracis-28398 322 31 , , PUNCT bracis-28398 322 32 muharemagic muharemagic PROPN bracis-28398 322 33 , , PUNCT bracis-28398 322 34 e. e. PROPN bracis-28398 322 35 : : PUNCT bracis-28398 322 36 deep deep ADJ bracis-28398 322 37 learning learning NOUN bracis-28398 322 38 applications application NOUN bracis-28398 322 39 and and CCONJ bracis-28398 322 40 challenges challenge NOUN bracis-28398 322 41 in in ADP bracis-28398 322 42 big big ADJ bracis-28398 322 43 data datum NOUN bracis-28398 322 44 analytics analytic NOUN bracis-28398 322 45 . . PUNCT bracis-28398 323 1 j. j. PROPN bracis-28398 323 2 big big PROPN bracis-28398 323 3 data data PROPN bracis-28398 323 4 2(1 2(1 NUM bracis-28398 323 5 ) ) PUNCT bracis-28398 323 6 , , PUNCT bracis-28398 323 7 1–21 1–21 PROPN bracis-28398 323 8 ( ( PUNCT bracis-28398 323 9 2015 2015 NUM bracis-28398 323 10 ) ) PUNCT bracis-28398 323 11 article article NOUN bracis-28398 323 12     SPACE bracis-28398 323 13 google google PROPN bracis-28398 323 14 scholar scholar NOUN bracis-28398 323 15     SPACE bracis-28398 323 16 download download NOUN bracis-28398 323 17 references reference NOUN bracis-28398 323 18 acknowledgements acknowledgement NOUN bracis-28398 323 19 this this DET bracis-28398 323 20 study study NOUN bracis-28398 323 21 was be AUX bracis-28398 323 22 financed finance VERB bracis-28398 323 23 in in ADP bracis-28398 323 24 part part NOUN bracis-28398 323 25 by by ADP bracis-28398 323 26 the the DET bracis-28398 323 27 coordenação coordenação PROPN bracis-28398 323 28 de de PROPN bracis-28398 323 29 aperfeiçoamento aperfeiçoamento PROPN bracis-28398 323 30 de de X bracis-28398 323 31 pessoal pessoal PROPN bracis-28398 323 32 de de PROPN bracis-28398 323 33 nível nível PROPN bracis-28398 323 34 superior superior PROPN bracis-28398 323 35 brasil brasil PROPN bracis-28398 323 36 ( ( PUNCT bracis-28398 323 37 capes cape NOUN bracis-28398 323 38 ) ) PUNCT bracis-28398 323 39 finance finance NOUN bracis-28398 323 40 code code NOUN bracis-28398 323 41 001 001 NOUN bracis-28398 323 42 . . PUNCT bracis-28398 324 1 the the DET bracis-28398 324 2 authors author NOUN bracis-28398 324 3 also also ADV bracis-28398 324 4 thank thank VERB bracis-28398 324 5 the the DET bracis-28398 324 6 financial financial ADJ bracis-28398 324 7 support support NOUN bracis-28398 324 8 of of ADP bracis-28398 324 9 fapesp fapesp NOUN bracis-28398 324 10 ( ( PUNCT bracis-28398 324 11 grant grant NOUN bracis-28398 324 12 2021/06870 2021/06870 NUM bracis-28398 324 13 - - SYM bracis-28398 324 14 3 3 NUM bracis-28398 324 15 ) ) PUNCT bracis-28398 324 16 and and CCONJ bracis-28398 324 17 cnpq cnpq NOUN bracis-28398 324 18 . . PUNCT bracis-28398 325 1 author author NOUN bracis-28398 325 2 information information NOUN bracis-28398 325 3 authors author NOUN bracis-28398 325 4 and and CCONJ bracis-28398 325 5 affiliations affiliation NOUN bracis-28398 325 6 instituto instituto PROPN bracis-28398 325 7 tecnológico tecnológico PROPN bracis-28398 325 8 de de PROPN bracis-28398 325 9 aeronáutica aeronáutica PROPN bracis-28398 325 10 , , PUNCT bracis-28398 325 11 praça praça PROPN bracis-28398 325 12 marechal marechal PROPN bracis-28398 325 13 eduardo eduardo PROPN bracis-28398 325 14 gomes gomes PROPN bracis-28398 325 15 , , PUNCT bracis-28398 325 16 são são PROPN bracis-28398 325 17 josé josé PROPN bracis-28398 325 18 dos dos PROPN bracis-28398 325 19 campos campos PROPN bracis-28398 325 20 , , PUNCT bracis-28398 325 21 brazil brazil PROPN bracis-28398 325 22 maria maria PROPN bracis-28398 325 23 gabriela gabriela PROPN bracis-28398 325 24 valeriano valeriano PROPN bracis-28398 325 25 , , PUNCT bracis-28398 325 26     SPACE bracis-28398 325 27 pedro pedro PROPN bracis-28398 325 28 yuri yuri PROPN bracis-28398 325 29 arbs arbs PROPN bracis-28398 325 30 paiva paiva PROPN bracis-28398 325 31     SPACE bracis-28398 325 32 & & CCONJ bracis-28398 325 33     SPACE bracis-28398 325 34 ana ana PROPN bracis-28398 325 35 carolina carolina PROPN bracis-28398 325 36 lorena lorena PROPN bracis-28398 325 37 universidade universidade PROPN bracis-28398 325 38 federal federal PROPN bracis-28398 325 39 de de PROPN bracis-28398 325 40 são são PROPN bracis-28398 325 41 paulo paulo PROPN bracis-28398 325 42 , , PUNCT bracis-28398 325 43 rua rua PROPN bracis-28398 325 44 botucatu botucatu PROPN bracis-28398 325 45 , , PUNCT bracis-28398 325 46 são são PROPN bracis-28398 325 47 paulo paulo PROPN bracis-28398 325 48 , , PUNCT bracis-28398 325 49 brazil brazil PROPN bracis-28398 325 50 maria maria PROPN bracis-28398 325 51 gabriela gabriela PROPN bracis-28398 325 52 valeriano valeriano PROPN bracis-28398 325 53     SPACE bracis-28398 325 54 & & CCONJ bracis-28398 325 55     SPACE bracis-28398 325 56 carlos carlos PROPN bracis-28398 325 57 roberto roberto PROPN bracis-28398 325 58 veiga veiga PROPN bracis-28398 325 59 kiffer kiffer PROPN bracis-28398 325 60 authors authors PROPN bracis-28398 325 61 maria maria PROPN bracis-28398 325 62 gabriela gabriela PROPN bracis-28398 325 63 valerianoview valerianoview PROPN bracis-28398 325 64 author author NOUN bracis-28398 325 65 publications publication NOUN bracis-28398 325 66 search search NOUN bracis-28398 325 67 author author NOUN bracis-28398 325 68 on on ADP bracis-28398 325 69 : : PUNCT bracis-28398 325 70 pubmed pubmed PROPN bracis-28398 325 71     SPACE bracis-28398 325 72 google google PROPN bracis-28398 325 73 scholar scholar NOUN bracis-28398 325 74 pedro pedro PROPN bracis-28398 325 75 yuri yuri PROPN bracis-28398 325 76 arbs arbs PROPN bracis-28398 325 77 paivaview paivaview PROPN bracis-28398 325 78 author author NOUN bracis-28398 325 79 publications publication NOUN bracis-28398 325 80 search search NOUN bracis-28398 325 81 author author NOUN bracis-28398 325 82 on on ADP bracis-28398 325 83 : : PUNCT bracis-28398 325 84 pubmed pubmed PROPN bracis-28398 325 85     SPACE bracis-28398 325 86 google google PROPN bracis-28398 325 87 scholar scholar PROPN bracis-28398 325 88 carlos carlos PROPN bracis-28398 325 89 roberto roberto PROPN bracis-28398 325 90 veiga veiga PROPN bracis-28398 325 91 kifferview kifferview PROPN bracis-28398 325 92 author author NOUN bracis-28398 325 93 publications publication NOUN bracis-28398 325 94 search search NOUN bracis-28398 325 95 author author NOUN bracis-28398 325 96 on on ADP bracis-28398 325 97 : : PUNCT bracis-28398 325 98 pubmed pubmed PROPN bracis-28398 325 99     SPACE bracis-28398 325 100 google google PROPN bracis-28398 325 101 scholar scholar PROPN bracis-28398 325 102 ana ana PROPN bracis-28398 325 103 carolina carolina PROPN bracis-28398 325 104 lorenaview lorenaview PROPN bracis-28398 325 105 author author NOUN bracis-28398 325 106 publications publication NOUN bracis-28398 325 107 search search NOUN bracis-28398 325 108 author author NOUN bracis-28398 325 109 on on ADP bracis-28398 325 110 : : PUNCT bracis-28398 325 111 pubmed pubmed PROPN bracis-28398 325 112     SPACE bracis-28398 325 113 google google PROPN bracis-28398 325 114 scholar scholar NOUN bracis-28398 325 115 corresponding correspond VERB bracis-28398 325 116 author author NOUN bracis-28398 325 117 correspondence correspondence NOUN bracis-28398 325 118 to to ADP bracis-28398 325 119 maria maria PROPN bracis-28398 325 120 gabriela gabriela PROPN bracis-28398 325 121 valeriano valeriano PROPN bracis-28398 325 122 . . PUNCT bracis-28398 326 1 editor editor NOUN bracis-28398 326 2 information information NOUN bracis-28398 326 3 editors editor NOUN bracis-28398 326 4 and and CCONJ bracis-28398 326 5 affiliations affiliation NOUN bracis-28398 326 6 federal federal PROPN bracis-28398 326 7 university university PROPN bracis-28398 326 8 of of ADP bracis-28398 326 9 são são PROPN bracis-28398 326 10 carlos carlos PROPN bracis-28398 326 11 , , PUNCT bracis-28398 326 12 são são PROPN bracis-28398 326 13 carlos carlos PROPN bracis-28398 326 14 , , PUNCT bracis-28398 326 15 brazil brazil PROPN bracis-28398 326 16 murilo murilo PROPN bracis-28398 326 17 c. c. PROPN bracis-28398 326 18 naldi naldi PROPN bracis-28398 326 19 centro centro PROPN bracis-28398 326 20 universitario universitario PROPN bracis-28398 326 21 da da PROPN bracis-28398 326 22 fei fei PROPN bracis-28398 326 23 , , PUNCT bracis-28398 326 24 são são PROPN bracis-28398 326 25 bernardo bernardo PROPN bracis-28398 326 26 do do AUX bracis-28398 326 27 campo campo PROPN bracis-28398 326 28 , , PUNCT bracis-28398 326 29 brazil brazil PROPN bracis-28398 326 30 reinaldo reinaldo PROPN bracis-28398 326 31 a. a. PROPN bracis-28398 326 32 c. c. PROPN bracis-28398 326 33 bianchi bianchi PROPN bracis-28398 326 34 rights right NOUN bracis-28398 326 35 and and CCONJ bracis-28398 326 36 permissions permission NOUN bracis-28398 326 37 reprints reprint NOUN bracis-28398 326 38 and and CCONJ bracis-28398 326 39 permissions permission VERB bracis-28398 326 40 copyright copyright NOUN bracis-28398 326 41 information information NOUN bracis-28398 326 42 © © ADP bracis-28398 326 43 2023 2023 NUM bracis-28398 326 44 the the DET bracis-28398 326 45 author(s author(s NOUN bracis-28398 326 46 ) ) PUNCT bracis-28398 326 47 , , PUNCT bracis-28398 326 48 under under ADP bracis-28398 326 49 exclusive exclusive ADJ bracis-28398 326 50 license license NOUN bracis-28398 326 51 to to ADP bracis-28398 326 52 springer springer NOUN bracis-28398 326 53 nature nature PROPN bracis-28398 326 54 switzerland switzerland PROPN bracis-28398 326 55 ag ag PROPN bracis-28398 326 56 about about ADP bracis-28398 326 57 this this DET bracis-28398 326 58 paper paper NOUN bracis-28398 326 59 cite cite VERB bracis-28398 326 60 this this DET bracis-28398 326 61 paper paper NOUN bracis-28398 326 62 valeriano valeriano NOUN bracis-28398 326 63 , , PUNCT bracis-28398 326 64 m.g m.g PROPN bracis-28398 326 65 . . PROPN bracis-28398 326 66 , , PUNCT bracis-28398 326 67 paiva paiva PROPN bracis-28398 326 68 , , PUNCT bracis-28398 326 69 p.y.a p.y.a NOUN bracis-28398 326 70 . . PUNCT bracis-28398 326 71 , , PUNCT bracis-28398 326 72 kiffer kiffer PROPN bracis-28398 326 73 , , PUNCT bracis-28398 326 74 c.r.v c.r.v PROPN bracis-28398 326 75 . . PUNCT bracis-28398 326 76 , , PUNCT bracis-28398 326 77 lorena lorena PROPN bracis-28398 326 78 , , PUNCT bracis-28398 326 79 a.c a.c PROPN bracis-28398 326 80 . . PROPN bracis-28398 326 81 ( ( PUNCT bracis-28398 326 82 2023 2023 NUM bracis-28398 326 83 ) ) PUNCT bracis-28398 326 84 . . PUNCT bracis-28398 327 1 a a DET bracis-28398 327 2 framework framework NOUN bracis-28398 327 3 for for ADP bracis-28398 327 4     SPACE bracis-28398 327 5 characterizing characterize VERB bracis-28398 327 6 what what PRON bracis-28398 327 7 makes make VERB bracis-28398 327 8 an an DET bracis-28398 327 9     SPACE bracis-28398 327 10 instance instance NOUN bracis-28398 327 11 hard hard ADV bracis-28398 327 12 to to PART bracis-28398 327 13     SPACE bracis-28398 327 14 classify classify VERB bracis-28398 327 15 . . PUNCT bracis-28398 328 1 in in ADP bracis-28398 328 2 : : PUNCT bracis-28398 328 3 naldi naldi PROPN bracis-28398 328 4 , , PUNCT bracis-28398 328 5 m.c m.c PROPN bracis-28398 328 6 . . PROPN bracis-28398 328 7 , , PUNCT bracis-28398 328 8 bianchi bianchi PROPN bracis-28398 328 9 , , PUNCT bracis-28398 328 10 r.a.c r.a.c ADP bracis-28398 328 11 . . PUNCT bracis-28398 328 12 ( ( PUNCT bracis-28398 328 13 eds ed NOUN bracis-28398 328 14 ) ) PUNCT bracis-28398 328 15 intelligent intelligent ADJ bracis-28398 328 16 systems system NOUN bracis-28398 328 17 . . PUNCT bracis-28398 328 18 bracis bracis PROPN bracis-28398 328 19 2023 2023 NUM bracis-28398 328 20 . . PUNCT bracis-28398 329 1 lecture lecture NOUN bracis-28398 329 2 notes note NOUN bracis-28398 329 3 in in ADP bracis-28398 329 4 computer computer NOUN bracis-28398 329 5 science science NOUN bracis-28398 329 6 ( ( PUNCT bracis-28398 329 7 ) ) PUNCT bracis-28398 329 8 , , PUNCT bracis-28398 329 9 vol vol NOUN bracis-28398 329 10 14196 14196 NUM bracis-28398 329 11 . . PUNCT bracis-28398 330 1 springer springer NOUN bracis-28398 330 2 , , PUNCT bracis-28398 330 3 cham cham PROPN bracis-28398 330 4 . . PUNCT bracis-28398 331 1 https://doi.org/10.1007/978-3-031-45389-2_24 https://doi.org/10.1007/978-3-031-45389-2_24 PROPN bracis-28398 331 2 download download NOUN bracis-28398 331 3 citation citation NOUN bracis-28398 331 4 .ris .ris PUNCT bracis-28398 331 5 .enw .enw PROPN bracis-28398 331 6 .bib .bib PUNCT bracis-28398 332 1 doi doi PROPN bracis-28398 332 2 : : PUNCT bracis-28398 332 3 https://doi.org/10.1007/978-3-031-45389-2_24 https://doi.org/10.1007/978-3-031-45389-2_24 PROPN bracis-28398 332 4 published publish VERB bracis-28398 332 5 : : PUNCT bracis-28398 332 6 12 12 NUM bracis-28398 332 7 october october PROPN bracis-28398 332 8 2023 2023 NUM bracis-28398 332 9 publisher publisher NOUN bracis-28398 332 10 name name NOUN bracis-28398 332 11 : : PUNCT bracis-28398 332 12 springer springer NOUN bracis-28398 332 13 , , PUNCT bracis-28398 332 14 cham cham PROPN bracis-28398 332 15 print print PROPN bracis-28398 332 16 isbn isbn PROPN bracis-28398 332 17 : : PUNCT bracis-28398 332 18 978 978 NUM bracis-28398 332 19 - - SYM bracis-28398 332 20 3 3 NUM bracis-28398 332 21 - - PUNCT bracis-28398 332 22 031 031 NUM bracis-28398 332 23 - - PUNCT bracis-28398 332 24 45388 45388 NUM bracis-28398 332 25 - - SYM bracis-28398 332 26 5 5 NUM bracis-28398 332 27 online online ADJ bracis-28398 332 28 isbn isbn NOUN bracis-28398 332 29 : : PUNCT bracis-28398 332 30 978 978 NUM bracis-28398 332 31 - - SYM bracis-28398 332 32 3 3 NUM bracis-28398 332 33 - - PUNCT bracis-28398 332 34 031 031 NUM bracis-28398 332 35 - - PUNCT bracis-28398 332 36 45389 45389 NUM bracis-28398 332 37 - - SYM bracis-28398 332 38 2 2 NUM bracis-28398 332 39 ebook ebook NOUN bracis-28398 332 40 packages package NOUN bracis-28398 332 41 : : PUNCT bracis-28398 332 42 computer computer NOUN bracis-28398 332 43 sciencecomputer sciencecomputer NOUN bracis-28398 332 44 science science NOUN bracis-28398 332 45 ( ( PUNCT bracis-28398 332 46 r0 r0 NOUN bracis-28398 332 47 ) ) PUNCT bracis-28398 332 48 share share VERB bracis-28398 332 49 this this DET bracis-28398 332 50 paper paper NOUN bracis-28398 332 51 anyone anyone PRON bracis-28398 332 52 you you PRON bracis-28398 332 53 share share VERB bracis-28398 332 54 the the DET bracis-28398 332 55 following follow VERB bracis-28398 332 56 link link NOUN bracis-28398 332 57 with with ADP bracis-28398 332 58 will will AUX bracis-28398 332 59 be be AUX bracis-28398 332 60 able able ADJ bracis-28398 332 61 to to PART bracis-28398 332 62 read read VERB bracis-28398 332 63 this this DET bracis-28398 332 64 content content NOUN bracis-28398 332 65 : : PUNCT bracis-28398 332 66 get get VERB bracis-28398 332 67 shareable shareable ADJ bracis-28398 332 68 linksorry linksorry NOUN bracis-28398 332 69 , , PUNCT bracis-28398 332 70 a a DET bracis-28398 332 71 shareable shareable ADJ bracis-28398 332 72 link link NOUN bracis-28398 332 73 is be AUX bracis-28398 332 74 not not PART bracis-28398 332 75 currently currently ADV bracis-28398 332 76 available available ADJ bracis-28398 332 77 for for ADP bracis-28398 332 78 this this DET bracis-28398 332 79 article article NOUN bracis-28398 332 80 . . PUNCT bracis-28398 333 1 copy copy VERB bracis-28398 333 2 shareable shareable ADJ bracis-28398 333 3 link link NOUN bracis-28398 333 4 to to PART bracis-28398 333 5 clipboard clipboard NOUN bracis-28398 333 6 provided provide VERB bracis-28398 333 7 by by ADP bracis-28398 333 8 the the DET bracis-28398 333 9 springer springer NOUN bracis-28398 333 10 nature nature PROPN bracis-28398 333 11 sharedit sharedit PROPN bracis-28398 333 12 content content NOUN bracis-28398 333 13 - - PUNCT bracis-28398 333 14 sharing share VERB bracis-28398 333 15 initiative initiative NOUN bracis-28398 333 16 keywords keyword VERB bracis-28398 333 17 explainability explainability NOUN bracis-28398 333 18 data data PROPN bracis-28398 333 19 - - PUNCT bracis-28398 333 20 center center NOUN bracis-28398 333 21 instance instance NOUN bracis-28398 333 22 - - PUNCT bracis-28398 333 23 hardness hardness NOUN bracis-28398 333 24 publish publish NOUN bracis-28398 333 25 with with ADP bracis-28398 333 26 us us PROPN bracis-28398 333 27 policies policy NOUN bracis-28398 333 28 and and CCONJ bracis-28398 333 29 ethics ethic NOUN bracis-28398 333 30 profiles profile NOUN bracis-28398 333 31 ana ana PROPN bracis-28398 333 32 carolina carolina PROPN bracis-28398 333 33 lorena lorena PROPN bracis-28398 333 34 view view PROPN bracis-28398 333 35 author author NOUN bracis-28398 333 36 profile profile NOUN bracis-28398 333 37 search search NOUN bracis-28398 333 38 search search NOUN bracis-28398 333 39 by by ADP bracis-28398 333 40 keyword keyword NOUN bracis-28398 333 41 or or CCONJ bracis-28398 333 42 author author NOUN bracis-28398 333 43 search search NOUN bracis-28398 333 44 navigation navigation NOUN bracis-28398 333 45 find find VERB bracis-28398 333 46 a a DET bracis-28398 333 47 journal journal NOUN bracis-28398 333 48 publish publish VERB bracis-28398 333 49 with with ADP bracis-28398 333 50 us we PRON bracis-28398 333 51 track track VERB bracis-28398 333 52 your your PRON bracis-28398 333 53 research research NOUN bracis-28398 333 54 discover discover VERB bracis-28398 333 55 content content NOUN bracis-28398 333 56 journals journal NOUN bracis-28398 333 57 a a DET bracis-28398 333 58 - - PUNCT bracis-28398 333 59 z z NOUN bracis-28398 333 60 books book NOUN bracis-28398 333 61 a a DET bracis-28398 333 62 - - PUNCT bracis-28398 333 63 z z NOUN bracis-28398 333 64 publish publish NOUN bracis-28398 333 65 with with ADP bracis-28398 333 66 us us PROPN bracis-28398 333 67 journal journal PROPN bracis-28398 333 68 finder finder PROPN bracis-28398 333 69 publish publish VERB bracis-28398 333 70 your your PRON bracis-28398 333 71 research research NOUN bracis-28398 333 72 language language NOUN bracis-28398 333 73 editing edit VERB bracis-28398 333 74 open open ADJ bracis-28398 333 75 access access NOUN bracis-28398 333 76 publishing publishing NOUN bracis-28398 333 77 products product NOUN bracis-28398 333 78 and and CCONJ bracis-28398 333 79 services service NOUN bracis-28398 333 80 our our PRON bracis-28398 333 81 products product NOUN bracis-28398 333 82 librarians librarian VERB bracis-28398 333 83 societies society NOUN bracis-28398 333 84 partners partner NOUN bracis-28398 333 85 and and CCONJ bracis-28398 333 86 advertisers advertiser NOUN bracis-28398 333 87 our our PRON bracis-28398 333 88 brands brand NOUN bracis-28398 333 89 springer springer NOUN bracis-28398 333 90 nature nature PROPN bracis-28398 333 91 portfolio portfolio PROPN bracis-28398 333 92 bmc bmc PROPN bracis-28398 333 93 palgrave palgrave PROPN bracis-28398 333 94 macmillan macmillan PROPN bracis-28398 333 95 apress apress PROPN bracis-28398 333 96 discover discover VERB bracis-28398 333 97 your your PRON bracis-28398 333 98 privacy privacy NOUN bracis-28398 333 99 choices choice NOUN bracis-28398 333 100 / / SYM bracis-28398 333 101 manage manage NOUN bracis-28398 333 102 cookies cookie NOUN bracis-28398 333 103 your your PRON bracis-28398 333 104 us us PROPN bracis-28398 334 1 state state NOUN bracis-28398 334 2 privacy privacy NOUN bracis-28398 334 3 rights right NOUN bracis-28398 334 4 accessibility accessibility NOUN bracis-28398 334 5 statement statement NOUN bracis-28398 334 6 terms term NOUN bracis-28398 334 7 and and CCONJ bracis-28398 334 8 conditions condition NOUN bracis-28398 334 9 privacy privacy NOUN bracis-28398 334 10 policy policy NOUN bracis-28398 334 11 help help NOUN bracis-28398 334 12 and and CCONJ bracis-28398 334 13 support support VERB bracis-28398 334 14 legal legal ADJ bracis-28398 334 15 notice notice NOUN bracis-28398 334 16 cancel cancel VERB bracis-28398 334 17 contracts contract NOUN bracis-28398 334 18 here here ADV bracis-28398 334 19 129.74.145.123 129.74.145.123 NUM bracis-28398 334 20 hesburgh hesburgh PROPN bracis-28398 334 21 library library PROPN bracis-28398 334 22 er er INTJ bracis-28398 334 23 unit unit NOUN bracis-28398 334 24 ( ( PUNCT bracis-28398 334 25 3005732405 3005732405 NUM bracis-28398 334 26 ) ) PUNCT bracis-28398 334 27 northeast northeast ADJ bracis-28398 334 28 research research NOUN bracis-28398 334 29 libraries library NOUN bracis-28398 334 30 ( ( PUNCT bracis-28398 334 31 nerl nerl PROPN bracis-28398 334 32 ) ) PUNCT bracis-28398 334 33 ( ( PUNCT bracis-28398 334 34 8200828607 8200828607 NUM bracis-28398 334 35 ) ) PUNCT bracis-28398 334 36 nerl nerl VERB bracis-28398 334 37 ta ta X bracis-28398 334 38 account account NOUN bracis-28398 334 39 ( ( PUNCT bracis-28398 334 40 3006206169 3006206169 NUM bracis-28398 334 41 ) ) PUNCT bracis-28398 334 42 university university NOUN bracis-28398 334 43 of of ADP bracis-28398 334 44 notre notre PROPN bracis-28398 334 45 dame dame PROPN bracis-28398 334 46 hesburgh hesburgh PROPN bracis-28398 334 47 library library NOUN bracis-28398 334 48 ( ( PUNCT bracis-28398 334 49 3000184373 3000184373 NUM bracis-28398 334 50 ) ) PUNCT bracis-28398 335 1 © © ADP bracis-28398 335 2 2025 2025 NUM bracis-28398 335 3 springer springer NOUN bracis-28398 335 4 nature nature NOUN