id sid tid token lemma pos bracis-28371 1 1 explainability explainability NOUN bracis-28371 1 2 of of ADP bracis-28371 1 3     SPACE bracis-28371 1 4 covid-19 covid-19 PROPN bracis-28371 1 5 classification classification NOUN bracis-28371 1 6 models model NOUN bracis-28371 1 7 using use VERB bracis-28371 1 8 dimensionality dimensionality NOUN bracis-28371 1 9 reduction reduction NOUN bracis-28371 1 10 of of ADP bracis-28371 1 11     SPACE bracis-28371 1 12 shap shap NOUN bracis-28371 1 13 values value NOUN bracis-28371 1 14 | | ADV bracis-28371 1 15 springer springer NOUN bracis-28371 1 16 nature nature PROPN bracis-28371 1 17 link link PROPN bracis-28371 1 18 ( ( PUNCT bracis-28371 1 19 formerly formerly ADV bracis-28371 1 20 springerlink springerlink NOUN bracis-28371 1 21 ) ) PUNCT bracis-28371 1 22 skip skip VERB bracis-28371 1 23 to to ADP bracis-28371 1 24 main main ADJ bracis-28371 1 25 content content NOUN bracis-28371 1 26 advertisement advertisement NOUN bracis-28371 1 27 log log NOUN bracis-28371 1 28 in in ADP bracis-28371 1 29 menu menu NOUN bracis-28371 1 30 find find VERB bracis-28371 1 31 a a DET bracis-28371 1 32 journal journal NOUN bracis-28371 1 33 publish publish VERB bracis-28371 1 34 with with ADP bracis-28371 1 35 us we PRON bracis-28371 1 36 track track VERB bracis-28371 1 37 your your PRON bracis-28371 1 38 research research NOUN bracis-28371 1 39 search search NOUN bracis-28371 1 40 cart cart NOUN bracis-28371 1 41 home home NOUN bracis-28371 1 42 intelligent intelligent ADJ bracis-28371 1 43 systems system NOUN bracis-28371 1 44 conference conference NOUN bracis-28371 1 45 paper paper NOUN bracis-28371 1 46 explainability explainability NOUN bracis-28371 1 47 of of ADP bracis-28371 1 48     SPACE bracis-28371 1 49 covid-19 covid-19 PROPN bracis-28371 1 50 classification classification NOUN bracis-28371 1 51 models model NOUN bracis-28371 1 52 using use VERB bracis-28371 1 53 dimensionality dimensionality NOUN bracis-28371 1 54 reduction reduction NOUN bracis-28371 1 55 of of ADP bracis-28371 1 56     SPACE bracis-28371 1 57 shap shap PROPN bracis-28371 1 58 values value NOUN bracis-28371 1 59 conference conference PROPN bracis-28371 1 60 paper paper NOUN bracis-28371 1 61 first first ADV bracis-28371 1 62 online online ADV bracis-28371 1 63 : : PUNCT bracis-28371 1 64 12 12 NUM bracis-28371 1 65 october october NOUN bracis-28371 1 66 2023 2023 NUM bracis-28371 1 67 pp pp ADP bracis-28371 1 68 415–430 415–430 NUM bracis-28371 1 69 cite cite VERB bracis-28371 1 70 this this DET bracis-28371 1 71 conference conference NOUN bracis-28371 1 72 paper paper NOUN bracis-28371 1 73 access access NOUN bracis-28371 1 74 provided provide VERB bracis-28371 1 75 by by ADP bracis-28371 1 76 university university PROPN bracis-28371 1 77 of of ADP bracis-28371 1 78 notre notre PROPN bracis-28371 1 79 dame dame PROPN bracis-28371 1 80 hesburgh hesburgh PROPN bracis-28371 1 81 library library PROPN bracis-28371 1 82 download download PROPN bracis-28371 1 83 book book NOUN bracis-28371 1 84 pdf pdf PROPN bracis-28371 1 85 download download NOUN bracis-28371 1 86 book book NOUN bracis-28371 1 87 epub epub PROPN bracis-28371 1 88 intelligent intelligent ADJ bracis-28371 1 89 systems system NOUN bracis-28371 1 90 ( ( PUNCT bracis-28371 1 91 bracis bracis NOUN bracis-28371 1 92 2023 2023 NUM bracis-28371 1 93 ) ) PUNCT bracis-28371 1 94 explainability explainability NOUN bracis-28371 1 95 of of ADP bracis-28371 1 96     SPACE bracis-28371 1 97 covid-19 covid-19 PROPN bracis-28371 1 98 classification classification NOUN bracis-28371 1 99 models model NOUN bracis-28371 1 100 using use VERB bracis-28371 1 101 dimensionality dimensionality NOUN bracis-28371 1 102 reduction reduction NOUN bracis-28371 1 103 of of ADP bracis-28371 1 104     SPACE bracis-28371 1 105 shap shap PROPN bracis-28371 1 106 values value NOUN bracis-28371 1 107 download download VERB bracis-28371 1 108 book book NOUN bracis-28371 1 109 pdf pdf PROPN bracis-28371 1 110 download download NOUN bracis-28371 1 111 book book PROPN bracis-28371 1 112 epub epub PROPN bracis-28371 1 113 daniel daniel PROPN bracis-28371 1 114 matheus matheus PROPN bracis-28371 1 115 kuhn9 kuhn9 PROPN bracis-28371 1 116 , , PUNCT bracis-28371 1 117 melina melina PROPN bracis-28371 1 118 silva silva PROPN bracis-28371 1 119 de de PROPN bracis-28371 1 120 loreto10 loreto10 PROPN bracis-28371 1 121 , , PUNCT bracis-28371 1 122 mariana mariana PROPN bracis-28371 1 123 recamonde recamonde PROPN bracis-28371 1 124 - - PUNCT bracis-28371 1 125 mendoza9 mendoza9 NOUN bracis-28371 1 126 , , PUNCT bracis-28371 1 127 joão joão PROPN bracis-28371 1 128 luiz luiz PROPN bracis-28371 1 129 dihl dihl PROPN bracis-28371 1 130 comba9 comba9 PROPN bracis-28371 1 131 & & CCONJ bracis-28371 1 132 … … PUNCT bracis-28371 1 133 viviane viviane PROPN bracis-28371 1 134 pereira pereira PROPN bracis-28371 1 135 moreira9 moreira9 PROPN bracis-28371 1 136     SPACE bracis-28371 1 137 show show NOUN bracis-28371 1 138 authors author NOUN bracis-28371 1 139 part part NOUN bracis-28371 1 140 of of ADP bracis-28371 1 141 the the DET bracis-28371 1 142 book book NOUN bracis-28371 1 143 series series NOUN bracis-28371 1 144 : : PUNCT bracis-28371 1 145 lecture lecture NOUN bracis-28371 1 146 notes note NOUN bracis-28371 1 147 in in ADP bracis-28371 1 148 computer computer NOUN bracis-28371 1 149 science science NOUN bracis-28371 1 150 ( ( PUNCT bracis-28371 1 151 ( ( PUNCT bracis-28371 1 152 lnai lnai ADJ bracis-28371 1 153 , , PUNCT bracis-28371 1 154 volume volume NOUN bracis-28371 1 155 14195 14195 NUM bracis-28371 1 156 ) ) PUNCT bracis-28371 1 157 ) ) PUNCT bracis-28371 1 158 included include VERB bracis-28371 1 159 in in ADP bracis-28371 1 160 the the DET bracis-28371 1 161 following follow VERB bracis-28371 1 162 conference conference NOUN bracis-28371 1 163 series series NOUN bracis-28371 1 164 : : PUNCT bracis-28371 1 165 brazilian brazilian ADJ bracis-28371 1 166 conference conference NOUN bracis-28371 1 167 on on ADP bracis-28371 1 168 intelligent intelligent ADJ bracis-28371 1 169 systems system NOUN bracis-28371 1 170 594 594 NUM bracis-28371 1 171 accesses access NOUN bracis-28371 1 172 abstract abstract VERB bracis-28371 1 173 the the DET bracis-28371 1 174 critical critical ADJ bracis-28371 1 175 scenario scenario NOUN bracis-28371 1 176 in in ADP bracis-28371 1 177 public public ADJ bracis-28371 1 178 health health NOUN bracis-28371 1 179 triggered trigger VERB bracis-28371 1 180 by by ADP bracis-28371 1 181 covid-19 covid-19 PROPN bracis-28371 1 182 intensified intensify VERB bracis-28371 1 183 the the DET bracis-28371 1 184 demand demand NOUN bracis-28371 1 185 for for ADP bracis-28371 1 186 predictive predictive ADJ bracis-28371 1 187 models model NOUN bracis-28371 1 188 to to PART bracis-28371 1 189 assist assist VERB bracis-28371 1 190 in in ADP bracis-28371 1 191 the the DET bracis-28371 1 192 diagnosis diagnosis NOUN bracis-28371 1 193 and and CCONJ bracis-28371 1 194 prognosis prognosis NOUN bracis-28371 1 195 of of ADP bracis-28371 1 196 patients patient NOUN bracis-28371 1 197 affected affect VERB bracis-28371 1 198 by by ADP bracis-28371 1 199 this this DET bracis-28371 1 200 disease disease NOUN bracis-28371 1 201 . . PUNCT bracis-28371 2 1 this this DET bracis-28371 2 2 work work NOUN bracis-28371 2 3 evaluates evaluate VERB bracis-28371 2 4 several several ADJ bracis-28371 2 5 machine machine NOUN bracis-28371 2 6 learning learn VERB bracis-28371 2 7 classifiers classifier NOUN bracis-28371 2 8 to to PART bracis-28371 2 9 predict predict VERB bracis-28371 2 10 the the DET bracis-28371 2 11 risk risk NOUN bracis-28371 2 12 of of ADP bracis-28371 2 13 covid-19 covid-19 PROPN bracis-28371 2 14 mortality mortality NOUN bracis-28371 2 15 based base VERB bracis-28371 2 16 on on ADP bracis-28371 2 17 information information NOUN bracis-28371 2 18 available available ADJ bracis-28371 2 19 at at ADP bracis-28371 2 20 the the DET bracis-28371 2 21 time time NOUN bracis-28371 2 22 of of ADP bracis-28371 2 23 admission admission NOUN bracis-28371 2 24 . . PUNCT bracis-28371 3 1 we we PRON bracis-28371 3 2 also also ADV bracis-28371 3 3 apply apply VERB bracis-28371 3 4 a a DET bracis-28371 3 5 visualization visualization NOUN bracis-28371 3 6 technique technique NOUN bracis-28371 3 7 based base VERB bracis-28371 3 8 on on ADP bracis-28371 3 9 a a DET bracis-28371 3 10 state state NOUN bracis-28371 3 11 - - PUNCT bracis-28371 3 12 of of ADP bracis-28371 3 13 - - PUNCT bracis-28371 3 14 the the DET bracis-28371 3 15 - - PUNCT bracis-28371 3 16 art art NOUN bracis-28371 3 17 explainability explainability NOUN bracis-28371 3 18 approach approach NOUN bracis-28371 3 19 which which PRON bracis-28371 3 20 , , PUNCT bracis-28371 3 21 combined combine VERB bracis-28371 3 22 with with ADP bracis-28371 3 23 a a DET bracis-28371 3 24 dimensionality dimensionality NOUN bracis-28371 3 25 reduction reduction NOUN bracis-28371 3 26 technique technique NOUN bracis-28371 3 27 , , PUNCT bracis-28371 3 28 allows allow VERB bracis-28371 3 29 drawing draw VERB bracis-28371 3 30 insights insight NOUN bracis-28371 3 31 into into ADP bracis-28371 3 32 the the DET bracis-28371 3 33 relationship relationship NOUN bracis-28371 3 34 between between ADP bracis-28371 3 35 the the DET bracis-28371 3 36 features feature NOUN bracis-28371 3 37 taken take VERB bracis-28371 3 38 into into ADP bracis-28371 3 39 account account NOUN bracis-28371 3 40 by by ADP bracis-28371 3 41 the the DET bracis-28371 3 42 classifiers classifier NOUN bracis-28371 3 43 in in ADP bracis-28371 3 44 their their PRON bracis-28371 3 45 predictions prediction NOUN bracis-28371 3 46 . . PUNCT bracis-28371 4 1 our our PRON bracis-28371 4 2 experiments experiment NOUN bracis-28371 4 3 on on ADP bracis-28371 4 4 two two NUM bracis-28371 4 5 real real ADJ bracis-28371 4 6 datasets dataset NOUN bracis-28371 4 7 showed show VERB bracis-28371 4 8 promising promising ADJ bracis-28371 4 9 results result NOUN bracis-28371 4 10 , , PUNCT bracis-28371 4 11 reaching reach VERB bracis-28371 4 12 a a DET bracis-28371 4 13 sensitivity sensitivity NOUN bracis-28371 4 14 of of ADP bracis-28371 4 15 up up ADP bracis-28371 4 16 to to PART bracis-28371 4 17 84 84 NUM bracis-28371 4 18 % % NOUN bracis-28371 4 19 and and CCONJ bracis-28371 4 20 an an DET bracis-28371 4 21 auroc auroc NOUN bracis-28371 4 22 of of ADP bracis-28371 4 23 92 92 NUM bracis-28371 4 24 % % NOUN bracis-28371 4 25 ( ( PUNCT bracis-28371 4 26 95 95 NUM bracis-28371 4 27 % % NOUN bracis-28371 4 28 ci ci NOUN bracis-28371 4 29 , , PUNCT bracis-28371 4 30 [ [ X bracis-28371 4 31 0.89–0.95 0.89–0.95 NOUN bracis-28371 4 32 ] ] NUM bracis-28371 4 33 ) ) PUNCT bracis-28371 4 34 . . PUNCT bracis-28371 5 1 access access NOUN bracis-28371 5 2 provided provide VERB bracis-28371 5 3 by by ADP bracis-28371 5 4 university university PROPN bracis-28371 5 5 of of ADP bracis-28371 5 6 notre notre PROPN bracis-28371 5 7 dame dame PROPN bracis-28371 5 8 hesburgh hesburgh PROPN bracis-28371 5 9 library library PROPN bracis-28371 5 10 . . PUNCT bracis-28371 6 1 download download PROPN bracis-28371 6 2 conference conference NOUN bracis-28371 6 3 paper paper NOUN bracis-28371 6 4 pdf pdf NOUN bracis-28371 6 5 similar similar ADJ bracis-28371 6 6 content content NOUN bracis-28371 6 7 being be AUX bracis-28371 6 8 viewed view VERB bracis-28371 6 9 by by ADP bracis-28371 6 10 others other NOUN bracis-28371 6 11 prognosis prognosis NOUN bracis-28371 6 12 patients patient NOUN bracis-28371 6 13 with with ADP bracis-28371 6 14 covid-19 covid-19 PROPN bracis-28371 6 15 using use VERB bracis-28371 6 16 deep deep ADJ bracis-28371 6 17 learning learning NOUN bracis-28371 6 18 article article NOUN bracis-28371 6 19 open open ADJ bracis-28371 6 20 access access NOUN bracis-28371 6 21 26 26 NUM bracis-28371 6 22 march march NOUN bracis-28371 6 23 2022 2022 NUM bracis-28371 6 24 improving improve VERB bracis-28371 6 25 cnn cnn PROPN bracis-28371 6 26 predictive predictive ADJ bracis-28371 6 27 accuracy accuracy NOUN bracis-28371 6 28 in in ADP bracis-28371 6 29 covid-19 covid-19 PROPN bracis-28371 6 30 health health NOUN bracis-28371 6 31 analytics analytic NOUN bracis-28371 6 32 article article NOUN bracis-28371 6 33 open open VERB bracis-28371 6 34 access access NOUN bracis-28371 6 35 14 14 NUM bracis-28371 6 36 august august PROPN bracis-28371 6 37 2025 2025 NUM bracis-28371 6 38 using use VERB bracis-28371 6 39 explainable explainable ADJ bracis-28371 6 40 machine machine NOUN bracis-28371 6 41 learning learn VERB bracis-28371 6 42 to to PART bracis-28371 6 43 characterise characterise VERB bracis-28371 6 44 data datum NOUN bracis-28371 6 45 drift drift NOUN bracis-28371 6 46 and and CCONJ bracis-28371 6 47 detect detect VERB bracis-28371 6 48 emergent emergent ADJ bracis-28371 6 49 health health NOUN bracis-28371 6 50 risks risk NOUN bracis-28371 6 51 for for ADP bracis-28371 6 52 emergency emergency NOUN bracis-28371 6 53 department department NOUN bracis-28371 6 54 admissions admission NOUN bracis-28371 6 55 during during ADP bracis-28371 6 56 covid-19 covid-19 PROPN bracis-28371 6 57 article article NOUN bracis-28371 6 58 open open VERB bracis-28371 6 59 access access NOUN bracis-28371 6 60 26 26 NUM bracis-28371 6 61 november november PROPN bracis-28371 6 62 2021 2021 NUM bracis-28371 6 63 explore explore VERB bracis-28371 6 64 related relate VERB bracis-28371 6 65 subjects subject NOUN bracis-28371 6 66 discover discover VERB bracis-28371 6 67 the the DET bracis-28371 6 68 latest late ADJ bracis-28371 6 69 articles article NOUN bracis-28371 6 70 , , PUNCT bracis-28371 6 71 books book NOUN bracis-28371 6 72 and and CCONJ bracis-28371 6 73 news news NOUN bracis-28371 6 74 in in ADP bracis-28371 6 75 related related ADJ bracis-28371 6 76 subjects subject NOUN bracis-28371 6 77 , , PUNCT bracis-28371 6 78 suggested suggest VERB bracis-28371 6 79 using use VERB bracis-28371 6 80 machine machine NOUN bracis-28371 6 81 learning learning NOUN bracis-28371 6 82 . . PUNCT bracis-28371 7 1 categorization categorization NOUN bracis-28371 7 2 data datum NOUN bracis-28371 7 3 science science PROPN bracis-28371 7 4 data data PROPN bracis-28371 7 5 analytics analytics PROPN bracis-28371 7 6 data data PROPN bracis-28371 7 7 mining mining NOUN bracis-28371 7 8 machine machine NOUN bracis-28371 7 9 learning learn VERB bracis-28371 7 10 statistical statistical ADJ bracis-28371 7 11 learning learn VERB bracis-28371 7 12 1 1 NUM bracis-28371 7 13 introduction introduction NOUN bracis-28371 7 14 the the DET bracis-28371 7 15 development development NOUN bracis-28371 7 16 of of ADP bracis-28371 7 17 predictive predictive ADJ bracis-28371 7 18 models model NOUN bracis-28371 7 19 for for ADP bracis-28371 7 20 the the DET bracis-28371 7 21 diagnosis diagnosis NOUN bracis-28371 7 22 and and CCONJ bracis-28371 7 23 prognosis prognosis NOUN bracis-28371 7 24 of of ADP bracis-28371 7 25 patients patient NOUN bracis-28371 7 26 has have AUX bracis-28371 7 27 been be AUX bracis-28371 7 28 recently recently ADV bracis-28371 7 29 investigated investigate VERB bracis-28371 7 30 by by ADP bracis-28371 7 31 numerous numerous ADJ bracis-28371 7 32 works work NOUN bracis-28371 7 33 [ [ X bracis-28371 7 34 17 17 NUM bracis-28371 7 35 ] ] PUNCT bracis-28371 7 36 . . PUNCT bracis-28371 8 1 the the DET bracis-28371 8 2 increasing increase VERB bracis-28371 8 3 interest interest NOUN bracis-28371 8 4 in in ADP bracis-28371 8 5 clinical clinical ADJ bracis-28371 8 6 predictive predictive ADJ bracis-28371 8 7 models model NOUN bracis-28371 8 8 may may AUX bracis-28371 8 9 be be AUX bracis-28371 8 10 explained explain VERB bracis-28371 8 11 by by ADP bracis-28371 8 12 factors factor NOUN bracis-28371 8 13 such such ADJ bracis-28371 8 14 as as ADP bracis-28371 8 15 the the DET bracis-28371 8 16 growing grow VERB bracis-28371 8 17 capacity capacity NOUN bracis-28371 8 18 of of ADP bracis-28371 8 19 data datum NOUN bracis-28371 8 20 collection collection NOUN bracis-28371 8 21 , , PUNCT bracis-28371 8 22 the the DET bracis-28371 8 23 positive positive ADJ bracis-28371 8 24 impact impact NOUN bracis-28371 8 25 on on ADP bracis-28371 8 26 decision decision NOUN bracis-28371 8 27 - - PUNCT bracis-28371 8 28 making make VERB bracis-28371 8 29 and and CCONJ bracis-28371 8 30 clinical clinical ADJ bracis-28371 8 31 care care NOUN bracis-28371 8 32 , , PUNCT bracis-28371 8 33 and and CCONJ bracis-28371 8 34 the the DET bracis-28371 8 35 great great ADJ bracis-28371 8 36 advances advance NOUN bracis-28371 8 37 in in ADP bracis-28371 8 38 machine machine NOUN bracis-28371 8 39 learning learning NOUN bracis-28371 8 40 ( ( PUNCT bracis-28371 8 41 ml ml NOUN bracis-28371 8 42 ) ) PUNCT bracis-28371 8 43 . . PUNCT bracis-28371 9 1 these these DET bracis-28371 9 2 models model NOUN bracis-28371 9 3 aim aim VERB bracis-28371 9 4 to to PART bracis-28371 9 5 assist assist VERB bracis-28371 9 6 in in ADP bracis-28371 9 7 the the DET bracis-28371 9 8 various various ADJ bracis-28371 9 9 challenges challenge NOUN bracis-28371 9 10 faced face VERB bracis-28371 9 11 by by ADP bracis-28371 9 12 health health NOUN bracis-28371 9 13 professionals professional NOUN bracis-28371 9 14 while while SCONJ bracis-28371 9 15 treating treat VERB bracis-28371 9 16 patients patient NOUN bracis-28371 9 17 , , PUNCT bracis-28371 9 18 optimizing optimize VERB bracis-28371 9 19 clinical clinical ADJ bracis-28371 9 20 management management NOUN bracis-28371 9 21 processes process NOUN bracis-28371 9 22 , , PUNCT bracis-28371 9 23 and and CCONJ bracis-28371 9 24 allocating allocate VERB bracis-28371 9 25 hospital hospital NOUN bracis-28371 9 26 resources resource NOUN bracis-28371 9 27 , , PUNCT bracis-28371 9 28 ultimately ultimately ADV bracis-28371 9 29 promoting promote VERB bracis-28371 9 30 better well ADJ bracis-28371 9 31 patient patient ADJ bracis-28371 9 32 care care NOUN bracis-28371 9 33 . . PUNCT bracis-28371 10 1 the the DET bracis-28371 10 2 covid-19 covid-19 PROPN bracis-28371 10 3 pandemic pandemic NOUN bracis-28371 10 4 was be AUX bracis-28371 10 5 a a DET bracis-28371 10 6 critical critical ADJ bracis-28371 10 7 public public ADJ bracis-28371 10 8 health health NOUN bracis-28371 10 9 emergency emergency NOUN bracis-28371 10 10 worldwide worldwide ADV bracis-28371 10 11 , , PUNCT bracis-28371 10 12 given give VERB bracis-28371 10 13 the the DET bracis-28371 10 14 sudden sudden ADJ bracis-28371 10 15 need need NOUN bracis-28371 10 16 for for ADP bracis-28371 10 17 many many ADJ bracis-28371 10 18 hospital hospital NOUN bracis-28371 10 19 beds bed NOUN bracis-28371 10 20 and and CCONJ bracis-28371 10 21 increased increase VERB bracis-28371 10 22 demand demand NOUN bracis-28371 10 23 for for ADP bracis-28371 10 24 medical medical ADJ bracis-28371 10 25 equipment equipment NOUN bracis-28371 10 26 and and CCONJ bracis-28371 10 27 health health NOUN bracis-28371 10 28 professionals professional NOUN bracis-28371 10 29 . . PUNCT bracis-28371 11 1 this this DET bracis-28371 11 2 scenario scenario NOUN bracis-28371 11 3 intensified intensify VERB bracis-28371 11 4 the the DET bracis-28371 11 5 demand demand NOUN bracis-28371 11 6 for for ADP bracis-28371 11 7 resources resource NOUN bracis-28371 11 8 and and CCONJ bracis-28371 11 9 , , PUNCT bracis-28371 11 10 consequently consequently ADV bracis-28371 11 11 , , PUNCT bracis-28371 11 12 the the DET bracis-28371 11 13 interest interest NOUN bracis-28371 11 14 and and CCONJ bracis-28371 11 15 need need VERB bracis-28371 11 16 in in ADP bracis-28371 11 17 using use VERB bracis-28371 11 18 predictive predictive ADJ bracis-28371 11 19 models model NOUN bracis-28371 11 20 to to PART bracis-28371 11 21 support support VERB bracis-28371 11 22 decision decision NOUN bracis-28371 11 23 - - PUNCT bracis-28371 11 24 making making NOUN bracis-28371 11 25 for for ADP bracis-28371 11 26 diagnosis diagnosis NOUN bracis-28371 11 27 and and CCONJ bracis-28371 11 28 prognosis prognosis NOUN bracis-28371 11 29 in in ADP bracis-28371 11 30 the the DET bracis-28371 11 31 context context NOUN bracis-28371 11 32 of of ADP bracis-28371 11 33 covid-19 covid-19 PROPN bracis-28371 11 34 . . PUNCT bracis-28371 12 1 as as SCONJ bracis-28371 12 2 discussed discuss VERB bracis-28371 12 3 by by ADP bracis-28371 12 4 previous previous ADJ bracis-28371 12 5 works work NOUN bracis-28371 12 6     SPACE bracis-28371 13 1 [ [ X bracis-28371 13 2 2 2 NUM bracis-28371 13 3 ] ] PUNCT bracis-28371 13 4 , , PUNCT bracis-28371 13 5 a a DET bracis-28371 13 6 considerable considerable ADJ bracis-28371 13 7 number number NOUN bracis-28371 13 8 of of ADP bracis-28371 13 9 papers paper NOUN bracis-28371 13 10 related relate VERB bracis-28371 13 11 to to ADP bracis-28371 13 12 covid-19 covid-19 PROPN bracis-28371 13 13 were be AUX bracis-28371 13 14 published publish VERB bracis-28371 13 15 in in ADP bracis-28371 13 16 a a DET bracis-28371 13 17 short short ADJ bracis-28371 13 18 period period NOUN bracis-28371 13 19 , , PUNCT bracis-28371 13 20 including include VERB bracis-28371 13 21 papers paper NOUN bracis-28371 13 22 proposing propose VERB bracis-28371 13 23 ml ml ADP bracis-28371 13 24 applications application NOUN bracis-28371 13 25 for for ADP bracis-28371 13 26 covid-19 covid-19 PROPN bracis-28371 13 27 prognosis prognosis NOUN bracis-28371 13 28 prediction prediction NOUN bracis-28371 13 29 . . PUNCT bracis-28371 14 1 among among ADP bracis-28371 14 2 the the DET bracis-28371 14 3 models model NOUN bracis-28371 14 4 for for ADP bracis-28371 14 5 prognosis prognosis NOUN bracis-28371 14 6 prediction prediction NOUN bracis-28371 14 7 , , PUNCT bracis-28371 14 8 we we PRON bracis-28371 14 9 can can AUX bracis-28371 14 10 cite cite VERB bracis-28371 14 11 predictive predictive ADJ bracis-28371 14 12 models model NOUN bracis-28371 14 13 that that PRON bracis-28371 14 14 aim aim VERB bracis-28371 14 15 to to PART bracis-28371 14 16 anticipate anticipate VERB bracis-28371 14 17 the the DET bracis-28371 14 18 patient patient NOUN bracis-28371 14 19 ’s ’s PART bracis-28371 14 20 outcome outcome NOUN bracis-28371 14 21 ( ( PUNCT bracis-28371 14 22 e.g. e.g. ADV bracis-28371 14 23 , , PUNCT bracis-28371 14 24 hospital hospital NOUN bracis-28371 14 25 discharge discharge NOUN bracis-28371 14 26 or or CCONJ bracis-28371 14 27 death death NOUN bracis-28371 14 28 ) ) PUNCT bracis-28371 14 29 , , PUNCT bracis-28371 14 30 the the DET bracis-28371 14 31 severity severity NOUN bracis-28371 14 32 of of ADP bracis-28371 14 33 the the DET bracis-28371 14 34 disease disease NOUN bracis-28371 14 35 , , PUNCT bracis-28371 14 36 the the DET bracis-28371 14 37 hospitalization hospitalization NOUN bracis-28371 14 38 length length NOUN bracis-28371 14 39 , , PUNCT bracis-28371 14 40 the the DET bracis-28371 14 41 evolution evolution NOUN bracis-28371 14 42 of of ADP bracis-28371 14 43 the the DET bracis-28371 14 44 patient patient NOUN bracis-28371 14 45 ’s ’s PART bracis-28371 14 46 clinical clinical ADJ bracis-28371 14 47 condition condition NOUN bracis-28371 14 48 during during ADP bracis-28371 14 49 hospitalization hospitalization NOUN bracis-28371 14 50 , , PUNCT bracis-28371 14 51 the the DET bracis-28371 14 52 need need NOUN bracis-28371 14 53 for for ADP bracis-28371 14 54 intensive intensive ADJ bracis-28371 14 55 care care NOUN bracis-28371 14 56 unit unit NOUN bracis-28371 14 57 ( ( PUNCT bracis-28371 14 58 icu icu PROPN bracis-28371 14 59 ) ) PUNCT bracis-28371 14 60 admission admission NOUN bracis-28371 14 61 or or CCONJ bracis-28371 14 62 interventions intervention NOUN bracis-28371 14 63 such such ADJ bracis-28371 14 64 as as ADP bracis-28371 14 65 intubation intubation NOUN bracis-28371 14 66 , , PUNCT bracis-28371 14 67 and and CCONJ bracis-28371 14 68 the the DET bracis-28371 14 69 development development NOUN bracis-28371 14 70 of of ADP bracis-28371 14 71 complications complication NOUN bracis-28371 14 72 ( ( PUNCT bracis-28371 14 73 e.g. e.g. ADV bracis-28371 14 74 , , PUNCT bracis-28371 14 75 cardiac cardiac ADJ bracis-28371 14 76 problems problem NOUN bracis-28371 14 77 , , PUNCT bracis-28371 14 78 thrombosis thrombosis NOUN bracis-28371 14 79 , , PUNCT bracis-28371 14 80 acute acute ADJ bracis-28371 14 81 respiratory respiratory ADJ bracis-28371 14 82 syndrome syndrome NOUN bracis-28371 14 83 ) ) PUNCT bracis-28371 14 84     SPACE bracis-28371 15 1 [ [ X bracis-28371 15 2 16 16 NUM bracis-28371 15 3 , , PUNCT bracis-28371 15 4 17 17 NUM bracis-28371 15 5 ] ] PUNCT bracis-28371 15 6 . . PUNCT bracis-28371 16 1 in in ADP bracis-28371 16 2 this this DET bracis-28371 16 3 work work NOUN bracis-28371 16 4 , , PUNCT bracis-28371 16 5 we we PRON bracis-28371 16 6 are be AUX bracis-28371 16 7 particularly particularly ADV bracis-28371 16 8 interested interested ADJ bracis-28371 16 9 in in ADP bracis-28371 16 10 mortality mortality NOUN bracis-28371 16 11 prediction prediction NOUN bracis-28371 16 12 models model NOUN bracis-28371 16 13 . . PUNCT bracis-28371 17 1 mortality mortality NOUN bracis-28371 17 2 prediction prediction NOUN bracis-28371 17 3 models model NOUN bracis-28371 17 4 help help VERB bracis-28371 17 5 recognize recognize VERB bracis-28371 17 6 patients patient NOUN bracis-28371 17 7 with with ADP bracis-28371 17 8 higher high ADJ bracis-28371 17 9 risks risk NOUN bracis-28371 17 10 of of ADP bracis-28371 17 11 poor poor ADJ bracis-28371 17 12 outcomes outcome NOUN bracis-28371 17 13 at at ADP bracis-28371 17 14 the the DET bracis-28371 17 15 time time NOUN bracis-28371 17 16 of of ADP bracis-28371 17 17 admission admission NOUN bracis-28371 17 18 or or CCONJ bracis-28371 17 19 during during ADP bracis-28371 17 20 hospitalization hospitalization NOUN bracis-28371 17 21 , , PUNCT bracis-28371 17 22 providing provide VERB bracis-28371 17 23 support support NOUN bracis-28371 17 24 to to ADP bracis-28371 17 25 more more ADV bracis-28371 17 26 effective effective ADJ bracis-28371 17 27 triage triage NOUN bracis-28371 17 28 and and CCONJ bracis-28371 17 29 treatment treatment NOUN bracis-28371 17 30 of of ADP bracis-28371 17 31 ill ill ADJ bracis-28371 17 32 patients patient NOUN bracis-28371 17 33 and and CCONJ bracis-28371 17 34 reducing reduce VERB bracis-28371 17 35 death death NOUN bracis-28371 17 36 rates rate NOUN bracis-28371 17 37 . . PUNCT bracis-28371 18 1 although although SCONJ bracis-28371 18 2 previous previous ADJ bracis-28371 18 3 studies study NOUN bracis-28371 18 4 have have AUX bracis-28371 18 5 shown show VERB bracis-28371 18 6 that that SCONJ bracis-28371 18 7 ml ml PROPN bracis-28371 18 8 algorithms algorithm NOUN bracis-28371 18 9 achieved achieve VERB bracis-28371 18 10 promising promise VERB bracis-28371 18 11 results result NOUN bracis-28371 18 12 in in ADP bracis-28371 18 13 the the DET bracis-28371 18 14 mortality mortality NOUN bracis-28371 18 15 prediction prediction NOUN bracis-28371 18 16 task task NOUN bracis-28371 18 17     SPACE bracis-28371 19 1 [ [ X bracis-28371 19 2 2 2 NUM bracis-28371 19 3 , , PUNCT bracis-28371 19 4 9 9 NUM bracis-28371 19 5 , , PUNCT bracis-28371 19 6 18 18 NUM bracis-28371 19 7 ] ] PUNCT bracis-28371 19 8 , , PUNCT bracis-28371 19 9 the the DET bracis-28371 19 10 lack lack NOUN bracis-28371 19 11 of of ADP bracis-28371 19 12 explainability explainability NOUN bracis-28371 19 13 of of ADP bracis-28371 19 14 the the DET bracis-28371 19 15 features feature NOUN bracis-28371 19 16 taken take VERB bracis-28371 19 17 into into ADP bracis-28371 19 18 account account NOUN bracis-28371 19 19 by by ADP bracis-28371 19 20 the the DET bracis-28371 19 21 predictive predictive ADJ bracis-28371 19 22 models model NOUN bracis-28371 19 23 has have AUX bracis-28371 19 24 been be AUX bracis-28371 19 25 pointed point VERB bracis-28371 19 26 out out ADP bracis-28371 19 27 as as ADP bracis-28371 19 28 a a DET bracis-28371 19 29 limiting limit VERB bracis-28371 19 30 factor factor NOUN bracis-28371 19 31 for for ADP bracis-28371 19 32 their their PRON bracis-28371 19 33 application application NOUN bracis-28371 19 34 in in ADP bracis-28371 19 35 the the DET bracis-28371 19 36 clinical clinical ADJ bracis-28371 19 37 routine routine NOUN bracis-28371 19 38     SPACE bracis-28371 20 1 [ [ X bracis-28371 20 2 9 9 NUM bracis-28371 20 3 ] ] PUNCT bracis-28371 20 4 . . PUNCT bracis-28371 21 1 this this DET bracis-28371 21 2 work work NOUN bracis-28371 21 3 aims aim VERB bracis-28371 21 4 to to PART bracis-28371 21 5 evaluate evaluate VERB bracis-28371 21 6 several several ADJ bracis-28371 21 7 ml ml NOUN bracis-28371 21 8 algorithms algorithm NOUN bracis-28371 21 9 for for ADP bracis-28371 21 10 predicting predict VERB bracis-28371 21 11 covid-19 covid-19 PROPN bracis-28371 21 12 mortality mortality NOUN bracis-28371 21 13 and and CCONJ bracis-28371 21 14 explore explore VERB bracis-28371 21 15 state state NOUN bracis-28371 21 16 - - PUNCT bracis-28371 21 17 of of ADP bracis-28371 21 18 - - PUNCT bracis-28371 21 19 the the DET bracis-28371 21 20 - - PUNCT bracis-28371 21 21 art art NOUN bracis-28371 21 22 explainability explainability NOUN bracis-28371 21 23 approaches approach VERB bracis-28371 21 24 to to PART bracis-28371 21 25 understand understand VERB bracis-28371 21 26 the the DET bracis-28371 21 27 factors factor NOUN bracis-28371 21 28 considered consider VERB bracis-28371 21 29 during during ADP bracis-28371 21 30 prediction prediction NOUN bracis-28371 21 31 . . PUNCT bracis-28371 22 1 we we PRON bracis-28371 22 2 developed develop VERB bracis-28371 22 3 a a DET bracis-28371 22 4 visualization visualization NOUN bracis-28371 22 5 technique technique NOUN bracis-28371 22 6 based base VERB bracis-28371 22 7 on on ADP bracis-28371 22 8 shap shap PROPN bracis-28371 22 9 values value NOUN bracis-28371 22 10     SPACE bracis-28371 23 1 [ [ X bracis-28371 23 2 11 11 NUM bracis-28371 23 3 ] ] PUNCT bracis-28371 23 4 combined combine VERB bracis-28371 23 5 with with ADP bracis-28371 23 6 a a DET bracis-28371 23 7 dimensionality dimensionality NOUN bracis-28371 23 8 reduction reduction NOUN bracis-28371 23 9 method method NOUN bracis-28371 23 10 that that PRON bracis-28371 23 11 draws draw VERB bracis-28371 23 12 insights insight NOUN bracis-28371 23 13 into into ADP bracis-28371 23 14 the the DET bracis-28371 23 15 relationship relationship NOUN bracis-28371 23 16 between between ADP bracis-28371 23 17 the the DET bracis-28371 23 18 features feature NOUN bracis-28371 23 19 taken take VERB bracis-28371 23 20 into into ADP bracis-28371 23 21 account account NOUN bracis-28371 23 22 by by ADP bracis-28371 23 23 the the DET bracis-28371 23 24 predictive predictive ADJ bracis-28371 23 25 models model NOUN bracis-28371 23 26 and and CCONJ bracis-28371 23 27 their their PRON bracis-28371 23 28 correct correct ADJ bracis-28371 23 29 and and CCONJ bracis-28371 23 30 incorrect incorrect ADJ bracis-28371 23 31 predictions prediction NOUN bracis-28371 23 32 . . PUNCT bracis-28371 24 1 to to PART bracis-28371 24 2 assess assess VERB bracis-28371 24 3 our our PRON bracis-28371 24 4 approach approach NOUN bracis-28371 24 5 , , PUNCT bracis-28371 24 6 we we PRON bracis-28371 24 7 conducted conduct VERB bracis-28371 24 8 a a DET bracis-28371 24 9 retrospective retrospective ADJ bracis-28371 24 10 study study NOUN bracis-28371 24 11 of of ADP bracis-28371 24 12 patients patient NOUN bracis-28371 24 13 diagnosed diagnose VERB bracis-28371 24 14 with with ADP bracis-28371 24 15 covid-19 covid-19 PROPN bracis-28371 24 16 , , PUNCT bracis-28371 24 17 analyzing analyze VERB bracis-28371 24 18 data datum NOUN bracis-28371 24 19 collected collect VERB bracis-28371 24 20 during during ADP bracis-28371 24 21 hospitalization hospitalization NOUN bracis-28371 24 22 in in ADP bracis-28371 24 23 two two NUM bracis-28371 24 24 hospitals hospital NOUN bracis-28371 24 25 . . PUNCT bracis-28371 25 1 we we PRON bracis-28371 25 2 also also ADV bracis-28371 25 3 analyze analyze VERB bracis-28371 25 4 the the DET bracis-28371 25 5 ability ability NOUN bracis-28371 25 6 of of ADP bracis-28371 25 7 classifiers classifier NOUN bracis-28371 25 8 to to PART bracis-28371 25 9 identify identify VERB bracis-28371 25 10 patients patient NOUN bracis-28371 25 11 with with ADP bracis-28371 25 12 a a DET bracis-28371 25 13 high high ADJ bracis-28371 25 14 probability probability NOUN bracis-28371 25 15 of of ADP bracis-28371 25 16 dying die VERB bracis-28371 25 17 at at ADP bracis-28371 25 18 different different ADJ bracis-28371 25 19 time time NOUN bracis-28371 25 20 intervals interval NOUN bracis-28371 25 21 . . PUNCT bracis-28371 26 1 our our PRON bracis-28371 26 2 main main ADJ bracis-28371 26 3 contributions contribution NOUN bracis-28371 26 4 can can AUX bracis-28371 26 5 be be AUX bracis-28371 26 6 summarized summarize VERB bracis-28371 26 7 as as SCONJ bracis-28371 26 8 follows follow VERB bracis-28371 26 9 : : PUNCT bracis-28371 26 10 ( ( PUNCT bracis-28371 26 11 i i NOUN bracis-28371 26 12 ) ) PUNCT bracis-28371 26 13 evaluation evaluation NOUN bracis-28371 26 14 of of ADP bracis-28371 26 15 six six NUM bracis-28371 26 16 ml ml NOUN bracis-28371 26 17 algorithms algorithm NOUN bracis-28371 26 18 for for ADP bracis-28371 26 19 the the DET bracis-28371 26 20 mortality mortality NOUN bracis-28371 26 21 prediction prediction NOUN bracis-28371 26 22 task task NOUN bracis-28371 26 23 ; ; PUNCT bracis-28371 26 24 ( ( PUNCT bracis-28371 26 25 ii ii NOUN bracis-28371 26 26 ) ) PUNCT bracis-28371 26 27 proposal proposal NOUN bracis-28371 26 28 of of ADP bracis-28371 26 29 a a DET bracis-28371 26 30 visualization visualization NOUN bracis-28371 26 31 technique technique NOUN bracis-28371 26 32 based base VERB bracis-28371 26 33 on on ADP bracis-28371 26 34 shap shap PROPN bracis-28371 26 35 values value NOUN bracis-28371 26 36 to to PART bracis-28371 26 37 assist assist VERB bracis-28371 26 38 in in ADP bracis-28371 26 39 explainability explainability NOUN bracis-28371 26 40 ; ; PUNCT bracis-28371 26 41 and and CCONJ bracis-28371 26 42 ( ( PUNCT bracis-28371 26 43 iii iii X bracis-28371 26 44 ) ) PUNCT bracis-28371 26 45 experiments experiment NOUN bracis-28371 26 46 with with ADP bracis-28371 26 47 real real ADJ bracis-28371 26 48 data datum NOUN bracis-28371 26 49 from from ADP bracis-28371 26 50 two two NUM bracis-28371 26 51 hospitals hospital NOUN bracis-28371 26 52 . . PUNCT bracis-28371 27 1 our our PRON bracis-28371 27 2 findings finding NOUN bracis-28371 27 3 showed show VERB bracis-28371 27 4 that that SCONJ bracis-28371 27 5 , , PUNCT bracis-28371 27 6 in in ADP bracis-28371 27 7 general general ADJ bracis-28371 27 8 , , PUNCT bracis-28371 27 9 deceased deceased ADJ bracis-28371 27 10 patients patient NOUN bracis-28371 27 11 were be AUX bracis-28371 27 12 older old ADJ bracis-28371 27 13 . . PUNCT bracis-28371 28 1 in in ADP bracis-28371 28 2 addition addition NOUN bracis-28371 28 3 , , PUNCT bracis-28371 28 4 age age NOUN bracis-28371 28 5 , , PUNCT bracis-28371 28 6 oxygen oxygen NOUN bracis-28371 28 7 saturation saturation NOUN bracis-28371 28 8 , , PUNCT bracis-28371 28 9 heart heart NOUN bracis-28371 28 10 disease disease NOUN bracis-28371 28 11 , , PUNCT bracis-28371 28 12 and and CCONJ bracis-28371 28 13 hemogram hemogram PROPN bracis-28371 28 14 information information NOUN bracis-28371 28 15 also also ADV bracis-28371 28 16 tended tend VERB bracis-28371 28 17 to to PART bracis-28371 28 18 have have VERB bracis-28371 28 19 high high ADJ bracis-28371 28 20 contributions contribution NOUN bracis-28371 28 21 in in ADP bracis-28371 28 22 the the DET bracis-28371 28 23 predictions prediction NOUN bracis-28371 28 24 . . PUNCT bracis-28371 29 1 logistic logistic ADJ bracis-28371 29 2 regression regression NOUN bracis-28371 29 3 , , PUNCT bracis-28371 29 4 adaboost adaboost ADV bracis-28371 29 5 , , PUNCT bracis-28371 29 6 and and CCONJ bracis-28371 29 7 xgboost xgboost ADJ bracis-28371 29 8 models model NOUN bracis-28371 29 9 recurrently recurrently ADV bracis-28371 29 10 achieved achieve VERB bracis-28371 29 11 the the DET bracis-28371 29 12 highest high ADJ bracis-28371 29 13 sensitivity sensitivity NOUN bracis-28371 29 14 scores score NOUN bracis-28371 29 15 among among ADP bracis-28371 29 16 the the DET bracis-28371 29 17 evaluated evaluated ADJ bracis-28371 29 18 classifiers classifier NOUN bracis-28371 29 19 . . PUNCT bracis-28371 30 1 2 2 NUM bracis-28371 30 2 related related ADJ bracis-28371 30 3 work work NOUN bracis-28371 30 4 this this DET bracis-28371 30 5 section section NOUN bracis-28371 30 6 discusses discuss VERB bracis-28371 30 7 existing existing ADJ bracis-28371 30 8 works work NOUN bracis-28371 30 9 that that PRON bracis-28371 30 10 addressed address VERB bracis-28371 30 11 mortality mortality NOUN bracis-28371 30 12 prediction prediction NOUN bracis-28371 30 13 for for ADP bracis-28371 30 14 covid-19 covid-19 PROPN bracis-28371 30 15 patients patient NOUN bracis-28371 30 16 based base VERB bracis-28371 30 17 on on ADP bracis-28371 30 18 the the DET bracis-28371 30 19 information information NOUN bracis-28371 30 20 available available ADJ bracis-28371 30 21 at at ADP bracis-28371 30 22 hospital hospital NOUN bracis-28371 30 23 admission admission NOUN bracis-28371 30 24 . . PUNCT bracis-28371 31 1 we we PRON bracis-28371 31 2 considered consider VERB bracis-28371 31 3 works work NOUN bracis-28371 31 4 evaluating evaluate VERB bracis-28371 31 5 classical classical ADJ bracis-28371 31 6 risk risk NOUN bracis-28371 31 7 scores score NOUN bracis-28371 31 8 ( ( PUNCT bracis-28371 31 9 including include VERB bracis-28371 31 10 those those PRON bracis-28371 31 11 designed design VERB bracis-28371 31 12 especially especially ADV bracis-28371 31 13 for for ADP bracis-28371 31 14 covid-19 covid-19 PROPN bracis-28371 31 15 patients patient NOUN bracis-28371 31 16 and and CCONJ bracis-28371 31 17 those those PRON bracis-28371 31 18 that that PRON bracis-28371 31 19 precede precede VERB bracis-28371 31 20 the the DET bracis-28371 31 21 pandemic pandemic NOUN bracis-28371 31 22 ) ) PUNCT bracis-28371 31 23 and and CCONJ bracis-28371 31 24 ml ml ADP bracis-28371 31 25 classifiers classifier NOUN bracis-28371 31 26 . . PUNCT bracis-28371 32 1 several several ADJ bracis-28371 32 2 techniques technique NOUN bracis-28371 32 3 have have AUX bracis-28371 32 4 been be AUX bracis-28371 32 5 employed employ VERB bracis-28371 32 6 to to PART bracis-28371 32 7 address address VERB bracis-28371 32 8 mortality mortality NOUN bracis-28371 32 9 prediction prediction NOUN bracis-28371 32 10 , , PUNCT bracis-28371 32 11 such such ADJ bracis-28371 32 12 as as ADP bracis-28371 32 13 using use VERB bracis-28371 32 14 risk risk NOUN bracis-28371 32 15 scores score NOUN bracis-28371 32 16 based base VERB bracis-28371 32 17 on on ADP bracis-28371 32 18 classical classical ADJ bracis-28371 32 19 statistical statistical ADJ bracis-28371 32 20 approaches approach NOUN bracis-28371 32 21 and and CCONJ bracis-28371 32 22 popular popular ADJ bracis-28371 32 23 classification classification NOUN bracis-28371 32 24 algorithms algorithm NOUN bracis-28371 32 25     SPACE bracis-28371 33 1 [ [ X bracis-28371 33 2 5 5 NUM bracis-28371 33 3 , , PUNCT bracis-28371 33 4 9 9 NUM bracis-28371 33 5 , , PUNCT bracis-28371 33 6 20 20 NUM bracis-28371 33 7 ] ] PUNCT bracis-28371 33 8 . . PUNCT bracis-28371 34 1 covino covino PROPN bracis-28371 34 2     SPACE bracis-28371 34 3 et et PROPN bracis-28371 34 4 al al PROPN bracis-28371 34 5 . . PUNCT bracis-28371 34 6     SPACE bracis-28371 35 1 [ [ X bracis-28371 35 2 5 5 NUM bracis-28371 35 3 ] ] PUNCT bracis-28371 35 4 evaluated evaluate VERB bracis-28371 35 5 six six NUM bracis-28371 35 6 physiological physiological ADJ bracis-28371 35 7 scoring scoring NOUN bracis-28371 35 8 systems system NOUN bracis-28371 35 9 recurrently recurrently ADV bracis-28371 35 10 used use VERB bracis-28371 35 11 as as ADP bracis-28371 35 12 early early ADJ bracis-28371 35 13 warning warning NOUN bracis-28371 35 14 scores score NOUN bracis-28371 35 15 for for ADP bracis-28371 35 16 predicting predict VERB bracis-28371 35 17 the the DET bracis-28371 35 18 risk risk NOUN bracis-28371 35 19 of of ADP bracis-28371 35 20 hospitalization hospitalization NOUN bracis-28371 35 21 in in ADP bracis-28371 35 22 an an DET bracis-28371 35 23 icu icu NOUN bracis-28371 35 24 and and CCONJ bracis-28371 35 25 the the DET bracis-28371 35 26 risk risk NOUN bracis-28371 35 27 of of ADP bracis-28371 35 28 mortality mortality NOUN bracis-28371 35 29 — — PUNCT bracis-28371 35 30 both both DET bracis-28371 35 31 tasks task NOUN bracis-28371 35 32 considering consider VERB bracis-28371 35 33 up up ADP bracis-28371 35 34 to to PART bracis-28371 35 35 seven seven NUM bracis-28371 35 36 days day NOUN bracis-28371 35 37 . . PUNCT bracis-28371 36 1 the the DET bracis-28371 36 2 rapid rapid ADJ bracis-28371 36 3 emergency emergency NOUN bracis-28371 36 4 medicine medicine NOUN bracis-28371 36 5 score score NOUN bracis-28371 36 6 obtained obtain VERB bracis-28371 36 7 the the DET bracis-28371 36 8 best good ADJ bracis-28371 36 9 results result NOUN bracis-28371 36 10 with with ADP bracis-28371 36 11 an an DET bracis-28371 36 12 auroc auroc NOUN bracis-28371 36 13 of of ADP bracis-28371 36 14 0.823 0.823 NUM bracis-28371 36 15 ( ( PUNCT bracis-28371 36 16 95 95 NUM bracis-28371 36 17 % % NOUN bracis-28371 36 18 ci ci NOUN bracis-28371 36 19 , , PUNCT bracis-28371 36 20 [ [ X bracis-28371 36 21 0.778–0.863 0.778–0.863 NOUN bracis-28371 36 22 ] ] X bracis-28371 36 23 ) ) PUNCT bracis-28371 36 24 . . PUNCT bracis-28371 37 1 zhao zhao PROPN bracis-28371 37 2     SPACE bracis-28371 37 3 et et PROPN bracis-28371 37 4 al al PROPN bracis-28371 37 5 . . PUNCT bracis-28371 37 6     SPACE bracis-28371 38 1 [ [ X bracis-28371 38 2 20 20 NUM bracis-28371 38 3 ] ] PUNCT bracis-28371 38 4 developed develop VERB bracis-28371 38 5 specific specific ADJ bracis-28371 38 6 risk risk NOUN bracis-28371 38 7 scores score NOUN bracis-28371 38 8 for for ADP bracis-28371 38 9 predicting predict VERB bracis-28371 38 10 mortality mortality NOUN bracis-28371 38 11 and and CCONJ bracis-28371 38 12 icu icu PROPN bracis-28371 38 13 admission admission NOUN bracis-28371 38 14 . . PUNCT bracis-28371 39 1 the the DET bracis-28371 39 2 scores score NOUN bracis-28371 39 3 were be AUX bracis-28371 39 4 developed develop VERB bracis-28371 39 5 based base VERB bracis-28371 39 6 on on ADP bracis-28371 39 7 logistic logistic ADJ bracis-28371 39 8 regression regression NOUN bracis-28371 39 9 and and CCONJ bracis-28371 39 10 reached reach VERB bracis-28371 39 11 an an DET bracis-28371 39 12 auroc auroc NOUN bracis-28371 39 13 of of ADP bracis-28371 39 14 0.83 0.83 NUM bracis-28371 39 15 ( ( PUNCT bracis-28371 39 16 95 95 NUM bracis-28371 39 17 % % NOUN bracis-28371 39 18 ci ci NOUN bracis-28371 39 19 , , PUNCT bracis-28371 40 1 [ [ X bracis-28371 40 2 0.73–0.92 0.73–0.92 NUM bracis-28371 40 3 ] ] PUNCT bracis-28371 40 4 ) ) PUNCT bracis-28371 40 5 . . PUNCT bracis-28371 41 1 another another DET bracis-28371 41 2 score score NOUN bracis-28371 41 3 specifically specifically ADV bracis-28371 41 4 designed design VERB bracis-28371 41 5 to to PART bracis-28371 41 6 predict predict VERB bracis-28371 41 7 mortality mortality NOUN bracis-28371 41 8 risk risk NOUN bracis-28371 41 9 was be AUX bracis-28371 41 10 the the DET bracis-28371 41 11 4c 4c NUM bracis-28371 41 12 mortality mortality NOUN bracis-28371 41 13 score score NOUN bracis-28371 41 14     SPACE bracis-28371 41 15 [ [ X bracis-28371 41 16 9 9 NUM bracis-28371 41 17 ] ] PUNCT bracis-28371 41 18 . . PUNCT bracis-28371 42 1 it it PRON bracis-28371 42 2 uses use VERB bracis-28371 42 3 generalized generalized ADJ bracis-28371 42 4 additive additive ADJ bracis-28371 42 5 models model NOUN bracis-28371 42 6 to to PART bracis-28371 42 7 categorize categorize VERB bracis-28371 42 8 continuous continuous ADJ bracis-28371 42 9 feature feature NOUN bracis-28371 42 10 values value NOUN bracis-28371 42 11 and and CCONJ bracis-28371 42 12 pre pre ADJ bracis-28371 42 13 - - ADJ bracis-28371 42 14 select select ADJ bracis-28371 42 15 predictors predictor NOUN bracis-28371 42 16 , , PUNCT bracis-28371 42 17 followed follow VERB bracis-28371 42 18 by by ADP bracis-28371 42 19 training train VERB bracis-28371 42 20 a a DET bracis-28371 42 21 logistic logistic ADJ bracis-28371 42 22 regression regression NOUN bracis-28371 42 23 model model NOUN bracis-28371 42 24 . . PUNCT bracis-28371 43 1 the the DET bracis-28371 43 2 4c 4c NUM bracis-28371 43 3 score score NOUN bracis-28371 43 4 yielded yield VERB bracis-28371 43 5 an an DET bracis-28371 43 6 auroc auroc NOUN bracis-28371 43 7 of of ADP bracis-28371 43 8 0.767 0.767 NUM bracis-28371 43 9 ( ( PUNCT bracis-28371 43 10 95 95 NUM bracis-28371 43 11 % % NOUN bracis-28371 43 12 ci ci NOUN bracis-28371 43 13 , , PUNCT bracis-28371 43 14 [ [ X bracis-28371 43 15 0.760–0.773 0.760–0.773 X bracis-28371 43 16 ] ] X bracis-28371 43 17 ) ) PUNCT bracis-28371 43 18 , , PUNCT bracis-28371 43 19 which which PRON bracis-28371 43 20 was be AUX bracis-28371 43 21 outperformed outperform VERB bracis-28371 43 22 by by ADP bracis-28371 43 23 the the DET bracis-28371 43 24 xgboost xgboost PROPN bracis-28371 43 25 algorithm algorithm NOUN bracis-28371 43 26 by by ADP bracis-28371 43 27 a a DET bracis-28371 43 28 small small ADJ bracis-28371 43 29 margin margin NOUN bracis-28371 43 30 . . PUNCT bracis-28371 44 1 still still ADV bracis-28371 44 2 , , PUNCT bracis-28371 44 3 the the DET bracis-28371 44 4 authors author NOUN bracis-28371 44 5 claim claim VERB bracis-28371 44 6 that that SCONJ bracis-28371 44 7 their their PRON bracis-28371 44 8 score score NOUN bracis-28371 44 9 is be AUX bracis-28371 44 10 preferable preferable ADJ bracis-28371 44 11 due due ADP bracis-28371 44 12 to to ADP bracis-28371 44 13 the the DET bracis-28371 44 14 difficulty difficulty NOUN bracis-28371 44 15 in in ADP bracis-28371 44 16 interpreting interpret VERB bracis-28371 44 17 the the DET bracis-28371 44 18 predictions prediction NOUN bracis-28371 44 19 made make VERB bracis-28371 44 20 by by ADP bracis-28371 44 21 xgboost xgboost PROPN bracis-28371 44 22 . . PUNCT bracis-28371 45 1 paiva paiva PROPN bracis-28371 45 2     SPACE bracis-28371 45 3 et et PROPN bracis-28371 45 4 al al PROPN bracis-28371 45 5 . . PUNCT bracis-28371 45 6     SPACE bracis-28371 46 1 [ [ X bracis-28371 46 2 13 13 NUM bracis-28371 46 3 ] ] PUNCT bracis-28371 46 4 also also ADV bracis-28371 46 5 evaluated evaluate VERB bracis-28371 46 6 classification classification NOUN bracis-28371 46 7 algorithms algorithm NOUN bracis-28371 46 8 for for ADP bracis-28371 46 9 predicting predict VERB bracis-28371 46 10 mortality mortality NOUN bracis-28371 46 11 , , PUNCT bracis-28371 46 12 including include VERB bracis-28371 46 13 a a DET bracis-28371 46 14 transformer transformer NOUN bracis-28371 46 15 ( ( PUNCT bracis-28371 46 16 fnet fnet NOUN bracis-28371 46 17 ) ) PUNCT bracis-28371 46 18 , , PUNCT bracis-28371 46 19 convolutional convolutional ADJ bracis-28371 46 20 neural neural ADJ bracis-28371 46 21 networks network NOUN bracis-28371 46 22 ( ( PUNCT bracis-28371 46 23 cnn cnn PROPN bracis-28371 46 24 ) ) PUNCT bracis-28371 46 25 , , PUNCT bracis-28371 46 26 boosting boost VERB bracis-28371 46 27 algorithms algorithm NOUN bracis-28371 46 28 ( ( PUNCT bracis-28371 46 29 lightgbm lightgbm ADJ bracis-28371 46 30 ) ) PUNCT bracis-28371 46 31 , , PUNCT bracis-28371 46 32 random random NOUN bracis-28371 46 33 - - PUNCT bracis-28371 46 34 forest forest NOUN bracis-28371 46 35 ( ( PUNCT bracis-28371 46 36 rf rf NOUN bracis-28371 46 37 ) ) PUNCT bracis-28371 46 38 , , PUNCT bracis-28371 46 39 support support VERB bracis-28371 46 40 vector vector NOUN bracis-28371 46 41 machines machine NOUN bracis-28371 46 42 ( ( PUNCT bracis-28371 46 43 svm svm PROPN bracis-28371 46 44 ) ) PUNCT bracis-28371 46 45 , , PUNCT bracis-28371 46 46 and and CCONJ bracis-28371 46 47 k k X bracis-28371 46 48 - - PUNCT bracis-28371 46 49 nearest near ADJ bracis-28371 46 50 - - PUNCT bracis-28371 46 51 neighbors neighbor NOUN bracis-28371 46 52 ( ( PUNCT bracis-28371 46 53 knn knn PROPN bracis-28371 46 54 ) ) PUNCT bracis-28371 46 55 . . PUNCT bracis-28371 47 1 in in ADP bracis-28371 47 2 addition addition NOUN bracis-28371 47 3 , , PUNCT bracis-28371 47 4 an an DET bracis-28371 47 5 ensemble ensemble ADJ bracis-28371 47 6 meta meta NOUN bracis-28371 47 7 - - PUNCT bracis-28371 47 8 model model NOUN bracis-28371 47 9 was be AUX bracis-28371 47 10 trained train VERB bracis-28371 47 11 to to PART bracis-28371 47 12 combine combine VERB bracis-28371 47 13 all all DET bracis-28371 47 14 the the DET bracis-28371 47 15 aforementioned aforementioned ADJ bracis-28371 47 16 classifiers classifier NOUN bracis-28371 47 17 . . PUNCT bracis-28371 48 1 the the DET bracis-28371 48 2 ensemble ensemble ADJ bracis-28371 48 3 model model NOUN bracis-28371 48 4 achieved achieve VERB bracis-28371 48 5 the the DET bracis-28371 48 6 best good ADJ bracis-28371 48 7 results result NOUN bracis-28371 48 8 , , PUNCT bracis-28371 48 9 with with ADP bracis-28371 48 10 an an DET bracis-28371 48 11 auroc auroc NOUN bracis-28371 48 12 of of ADP bracis-28371 48 13 0.871 0.871 NUM bracis-28371 48 14 ( ( PUNCT bracis-28371 48 15 95 95 NUM bracis-28371 48 16 % % NOUN bracis-28371 48 17 ci ci NOUN bracis-28371 49 1 [ [ X bracis-28371 49 2 0.864–0.878 0.864–0.878 PROPN bracis-28371 49 3 ] ] X bracis-28371 49 4 ) ) PUNCT bracis-28371 49 5 . . PUNCT bracis-28371 50 1 to to PART bracis-28371 50 2 explain explain VERB bracis-28371 50 3 the the DET bracis-28371 50 4 predictive predictive ADJ bracis-28371 50 5 models model NOUN bracis-28371 50 6 , , PUNCT bracis-28371 50 7 the the DET bracis-28371 50 8 study study NOUN bracis-28371 50 9 contemplated contemplate VERB bracis-28371 50 10 the the DET bracis-28371 50 11 application application NOUN bracis-28371 50 12 of of ADP bracis-28371 50 13 shapley shapley ADJ bracis-28371 50 14 additive additive ADJ bracis-28371 50 15 explanations explanation NOUN bracis-28371 50 16 ( ( PUNCT bracis-28371 50 17 shap shap NOUN bracis-28371 50 18 ) ) PUNCT bracis-28371 50 19     SPACE bracis-28371 50 20 et et PROPN bracis-28371 50 21 al al PROPN bracis-28371 50 22 . . PUNCT bracis-28371 50 23     SPACE bracis-28371 51 1 [ [ X bracis-28371 51 2 11 11 NUM bracis-28371 51 3 ] ] PUNCT bracis-28371 51 4 . . PUNCT bracis-28371 52 1 araújo araújo PROPN bracis-28371 52 2     SPACE bracis-28371 52 3 et et PROPN bracis-28371 52 4 al al PROPN bracis-28371 52 5 . . PUNCT bracis-28371 52 6     SPACE bracis-28371 53 1 [ [ X bracis-28371 53 2 3 3 NUM bracis-28371 53 3 ] ] PUNCT bracis-28371 53 4 developed develop VERB bracis-28371 53 5 risk risk NOUN bracis-28371 53 6 models model NOUN bracis-28371 53 7 to to PART bracis-28371 53 8 predict predict VERB bracis-28371 53 9 the the DET bracis-28371 53 10 probability probability NOUN bracis-28371 53 11 of of ADP bracis-28371 53 12 covid-19 covid-19 PROPN bracis-28371 53 13 severity severity NOUN bracis-28371 53 14 . . PUNCT bracis-28371 54 1 to to PART bracis-28371 54 2 perform perform VERB bracis-28371 54 3 the the DET bracis-28371 54 4 study study NOUN bracis-28371 54 5 , , PUNCT bracis-28371 54 6 the the DET bracis-28371 54 7 authors author NOUN bracis-28371 54 8 collected collect VERB bracis-28371 54 9 blood blood NOUN bracis-28371 54 10 biomarkers biomarker NOUN bracis-28371 54 11 data datum NOUN bracis-28371 54 12 from from ADP bracis-28371 54 13 covid-19 covid-19 PROPN bracis-28371 54 14 patients patient NOUN bracis-28371 54 15 . . PUNCT bracis-28371 55 1 the the DET bracis-28371 55 2 models model NOUN bracis-28371 55 3 were be AUX bracis-28371 55 4 developed develop VERB bracis-28371 55 5 with with ADP bracis-28371 55 6 a a DET bracis-28371 55 7 lightgbm lightgbm ADJ bracis-28371 55 8 - - PUNCT bracis-28371 55 9 based base VERB bracis-28371 55 10 implementation implementation NOUN bracis-28371 55 11 , , PUNCT bracis-28371 55 12 which which PRON bracis-28371 55 13 achieved achieve VERB bracis-28371 55 14 an an DET bracis-28371 55 15 average average ADJ bracis-28371 55 16 auroc auroc NOUN bracis-28371 55 17 of of ADP bracis-28371 55 18 0.91 0.91 NUM bracis-28371 55 19 ± ± NUM bracis-28371 55 20 0.01 0.01 NUM bracis-28371 55 21 for for ADP bracis-28371 55 22 mortality mortality NOUN bracis-28371 55 23 prediction prediction NOUN bracis-28371 55 24 . . PUNCT bracis-28371 56 1 the the DET bracis-28371 56 2 authors author NOUN bracis-28371 56 3 of of ADP bracis-28371 56 4 this this DET bracis-28371 56 5 work work NOUN bracis-28371 56 6 apply apply VERB bracis-28371 56 7 a a DET bracis-28371 56 8 visualization visualization NOUN bracis-28371 56 9 technique technique NOUN bracis-28371 56 10 similar similar ADJ bracis-28371 56 11 to to ADP bracis-28371 56 12 the the DET bracis-28371 56 13 one one NOUN bracis-28371 56 14 proposed propose VERB bracis-28371 56 15 in in ADP bracis-28371 56 16 our our PRON bracis-28371 56 17 work work NOUN bracis-28371 56 18 . . PUNCT bracis-28371 57 1 the the DET bracis-28371 57 2 main main ADJ bracis-28371 57 3 difference difference NOUN bracis-28371 57 4 is be AUX bracis-28371 57 5 that that SCONJ bracis-28371 57 6 they they PRON bracis-28371 57 7 use use VERB bracis-28371 57 8 it it PRON bracis-28371 57 9 only only ADV bracis-28371 57 10 to to PART bracis-28371 57 11 assess assess VERB bracis-28371 57 12 the the DET bracis-28371 57 13 model model NOUN bracis-28371 57 14 ’s ’s PART bracis-28371 57 15 ability ability NOUN bracis-28371 57 16 to to PART bracis-28371 57 17 separate separate VERB bracis-28371 57 18 the the DET bracis-28371 57 19 targeted target VERB bracis-28371 57 20 classes class NOUN bracis-28371 57 21 . . PUNCT bracis-28371 58 1 conversely conversely ADV bracis-28371 58 2 , , PUNCT bracis-28371 58 3 in in ADP bracis-28371 58 4 our our PRON bracis-28371 58 5 work work NOUN bracis-28371 58 6 , , PUNCT bracis-28371 58 7 we we PRON bracis-28371 58 8 use use VERB bracis-28371 58 9 visualization visualization NOUN bracis-28371 58 10 to to PART bracis-28371 58 11 analyze analyze VERB bracis-28371 58 12 the the DET bracis-28371 58 13 relationship relationship NOUN bracis-28371 58 14 between between ADP bracis-28371 58 15 the the DET bracis-28371 58 16 features feature NOUN bracis-28371 58 17 taken take VERB bracis-28371 58 18 into into ADP bracis-28371 58 19 account account NOUN bracis-28371 58 20 by by ADP bracis-28371 58 21 the the DET bracis-28371 58 22 classifiers classifier NOUN bracis-28371 58 23 and and CCONJ bracis-28371 58 24 their their PRON bracis-28371 58 25 correct correct ADJ bracis-28371 58 26 and and CCONJ bracis-28371 58 27 incorrect incorrect ADJ bracis-28371 58 28 predictions prediction NOUN bracis-28371 58 29 . . PUNCT bracis-28371 59 1 the the DET bracis-28371 59 2 datasets dataset NOUN bracis-28371 59 3 used use VERB bracis-28371 59 4 in in ADP bracis-28371 59 5 the the DET bracis-28371 59 6 studies study NOUN bracis-28371 59 7 typically typically ADV bracis-28371 59 8 range range VERB bracis-28371 59 9 from from ADP bracis-28371 59 10 a a DET bracis-28371 59 11 single single ADJ bracis-28371 59 12 hospital hospital NOUN bracis-28371 59 13 unit unit NOUN bracis-28371 59 14 with with ADP bracis-28371 59 15 a a DET bracis-28371 59 16 few few ADJ bracis-28371 59 17 hundred hundred NUM bracis-28371 59 18 patients patient NOUN bracis-28371 59 19 to to ADP bracis-28371 59 20 datasets dataset NOUN bracis-28371 59 21 comprising comprise VERB bracis-28371 59 22 thousands thousand NOUN bracis-28371 59 23 of of ADP bracis-28371 59 24 hospitalization hospitalization NOUN bracis-28371 59 25 records record NOUN bracis-28371 59 26 from from ADP bracis-28371 59 27 hundreds hundred NOUN bracis-28371 59 28 of of ADP bracis-28371 59 29 hospital hospital NOUN bracis-28371 59 30 units unit NOUN bracis-28371 59 31 in in ADP bracis-28371 59 32 several several ADJ bracis-28371 59 33 countries country NOUN bracis-28371 59 34 . . PUNCT bracis-28371 60 1 covino covino PROPN bracis-28371 60 2     SPACE bracis-28371 60 3 et et PROPN bracis-28371 60 4 al al PROPN bracis-28371 60 5 . . PUNCT bracis-28371 60 6     SPACE bracis-28371 61 1 [ [ X bracis-28371 61 2 5 5 NUM bracis-28371 61 3 ] ] PUNCT bracis-28371 61 4 and and CCONJ bracis-28371 61 5 zhao zhao PROPN bracis-28371 61 6     SPACE bracis-28371 61 7 et et PROPN bracis-28371 61 8 al al PROPN bracis-28371 61 9 . . PUNCT bracis-28371 61 10     SPACE bracis-28371 62 1 [ [ X bracis-28371 62 2 20 20 NUM bracis-28371 62 3 ] ] PUNCT bracis-28371 62 4 carried carry VERB bracis-28371 62 5 out out ADP bracis-28371 62 6 their their PRON bracis-28371 62 7 studies study NOUN bracis-28371 62 8 based base VERB bracis-28371 62 9 on on ADP bracis-28371 62 10 datasets dataset NOUN bracis-28371 62 11 containing contain VERB bracis-28371 62 12 334 334 NUM bracis-28371 62 13 and and CCONJ bracis-28371 62 14 641 641 NUM bracis-28371 62 15 patients patient NOUN bracis-28371 62 16 , , PUNCT bracis-28371 62 17 respectively respectively ADV bracis-28371 62 18 , , PUNCT bracis-28371 62 19 from from ADP bracis-28371 62 20 a a DET bracis-28371 62 21 single single ADJ bracis-28371 62 22 hospital hospital NOUN bracis-28371 62 23 unit unit NOUN bracis-28371 62 24 . . PUNCT bracis-28371 63 1 paiva paiva PROPN bracis-28371 63 2     SPACE bracis-28371 63 3 et et PROPN bracis-28371 63 4 al al PROPN bracis-28371 63 5 . . PUNCT bracis-28371 63 6     SPACE bracis-28371 64 1 [ [ X bracis-28371 64 2 13 13 NUM bracis-28371 64 3 ] ] PUNCT bracis-28371 64 4 used use VERB bracis-28371 64 5 a a DET bracis-28371 64 6 dataset dataset NOUN bracis-28371 64 7 with with ADP bracis-28371 64 8 5,032 5,032 NUM bracis-28371 64 9 patients patient NOUN bracis-28371 64 10 from from ADP bracis-28371 64 11 36 36 NUM bracis-28371 64 12 hospitals hospital NOUN bracis-28371 64 13 in in ADP bracis-28371 64 14 17 17 NUM bracis-28371 64 15 cities city NOUN bracis-28371 64 16 in in ADP bracis-28371 64 17 5 5 NUM bracis-28371 64 18 brazilian brazilian ADJ bracis-28371 64 19 states state NOUN bracis-28371 64 20 , , PUNCT bracis-28371 64 21 and and CCONJ bracis-28371 64 22 araújo araújo PROPN bracis-28371 64 23     SPACE bracis-28371 64 24 et et PROPN bracis-28371 64 25 al al PROPN bracis-28371 64 26 . . PUNCT bracis-28371 64 27     SPACE bracis-28371 65 1 [ [ X bracis-28371 65 2 3 3 X bracis-28371 65 3 ] ] PUNCT bracis-28371 65 4 used use VERB bracis-28371 65 5 a a DET bracis-28371 65 6 dataset dataset NOUN bracis-28371 65 7 with with ADP bracis-28371 65 8 6,979 6,979 NUM bracis-28371 65 9 patients patient NOUN bracis-28371 65 10 from from ADP bracis-28371 65 11 two two NUM bracis-28371 65 12 brazilian brazilian ADJ bracis-28371 65 13 hospitals hospital NOUN bracis-28371 65 14 . . PUNCT bracis-28371 66 1 to to PART bracis-28371 66 2 develop develop VERB bracis-28371 66 3 the the DET bracis-28371 66 4 4c 4c NUM bracis-28371 66 5 mortality mortality NOUN bracis-28371 66 6 score score NOUN bracis-28371 66 7 , , PUNCT bracis-28371 66 8 knight knight PROPN bracis-28371 66 9     SPACE bracis-28371 66 10 et et PROPN bracis-28371 66 11 al al PROPN bracis-28371 66 12 . . PUNCT bracis-28371 66 13     SPACE bracis-28371 67 1 [ [ X bracis-28371 67 2 9 9 NUM bracis-28371 67 3 ] ] PUNCT bracis-28371 67 4 used use VERB bracis-28371 67 5 a a DET bracis-28371 67 6 large large ADJ bracis-28371 67 7 dataset dataset NOUN bracis-28371 67 8 with with ADP bracis-28371 67 9 57,824 57,824 NUM bracis-28371 67 10 patients patient NOUN bracis-28371 67 11 collected collect VERB bracis-28371 67 12 from from ADP bracis-28371 67 13 260 260 NUM bracis-28371 67 14 hospitals hospital NOUN bracis-28371 67 15 in in ADP bracis-28371 67 16 england england PROPN bracis-28371 67 17 , , PUNCT bracis-28371 67 18 scotland scotland PROPN bracis-28371 67 19 , , PUNCT bracis-28371 67 20 and and CCONJ bracis-28371 67 21 wales wale NOUN bracis-28371 67 22 . . PUNCT bracis-28371 68 1 although although SCONJ bracis-28371 68 2 the the DET bracis-28371 68 3 datasets dataset NOUN bracis-28371 68 4 used use VERB bracis-28371 68 5 in in ADP bracis-28371 68 6 the the DET bracis-28371 68 7 different different ADJ bracis-28371 68 8 studies study NOUN bracis-28371 68 9 do do AUX bracis-28371 68 10 not not PART bracis-28371 68 11 contain contain VERB bracis-28371 68 12 the the DET bracis-28371 68 13 same same ADJ bracis-28371 68 14 features feature NOUN bracis-28371 68 15 , , PUNCT bracis-28371 68 16 there there PRON bracis-28371 68 17 is be VERB bracis-28371 68 18 a a DET bracis-28371 68 19 recurrence recurrence NOUN bracis-28371 68 20 among among ADP bracis-28371 68 21 the the DET bracis-28371 68 22 features feature NOUN bracis-28371 68 23 considered consider VERB bracis-28371 68 24 most most ADV bracis-28371 68 25 relevant relevant ADJ bracis-28371 68 26 by by ADP bracis-28371 68 27 the the DET bracis-28371 68 28 predictive predictive ADJ bracis-28371 68 29 models model NOUN bracis-28371 68 30 . . PUNCT bracis-28371 69 1 age age NOUN bracis-28371 69 2 is be AUX bracis-28371 69 3 the the DET bracis-28371 69 4 most most ADV bracis-28371 69 5 frequently frequently ADV bracis-28371 69 6 cited cite VERB bracis-28371 69 7 variable variable NOUN bracis-28371 69 8 . . PUNCT bracis-28371 70 1 other other ADJ bracis-28371 70 2 features feature NOUN bracis-28371 70 3 include include VERB bracis-28371 70 4 oxygen oxygen NOUN bracis-28371 70 5 saturation saturation NOUN bracis-28371 70 6 , , PUNCT bracis-28371 70 7 respiratory respiratory ADJ bracis-28371 70 8 rate rate NOUN bracis-28371 70 9 , , PUNCT bracis-28371 70 10 comorbidities comorbiditie NOUN bracis-28371 70 11 , , PUNCT bracis-28371 70 12 heart heart NOUN bracis-28371 70 13 failure failure NOUN bracis-28371 70 14 , , PUNCT bracis-28371 70 15 lactate lactate NOUN bracis-28371 70 16 dehydrogenase dehydrogenase NOUN bracis-28371 70 17 ( ( PUNCT bracis-28371 70 18 ldh ldh PROPN bracis-28371 70 19 ) ) PUNCT bracis-28371 70 20 , , PUNCT bracis-28371 70 21 c c NOUN bracis-28371 70 22 - - PUNCT bracis-28371 70 23 reactive reactive ADJ bracis-28371 70 24 protein protein NOUN bracis-28371 70 25 ( ( PUNCT bracis-28371 70 26 crp crp PROPN bracis-28371 70 27 ) ) PUNCT bracis-28371 70 28 , , PUNCT bracis-28371 70 29 and and CCONJ bracis-28371 70 30 consciousness consciousness NOUN bracis-28371 70 31 level level NOUN bracis-28371 70 32 metrics metric NOUN bracis-28371 70 33 ( ( PUNCT bracis-28371 70 34 e.g. e.g. ADV bracis-28371 70 35 , , PUNCT bracis-28371 70 36 glasgow glasgow ADJ bracis-28371 70 37 coma coma NOUN bracis-28371 70 38 scale scale NOUN bracis-28371 70 39 ) ) PUNCT bracis-28371 70 40 . . PUNCT bracis-28371 71 1 our our PRON bracis-28371 71 2 study study NOUN bracis-28371 71 3 complements complement VERB bracis-28371 71 4 the the DET bracis-28371 71 5 aforementioned aforementione VERB bracis-28371 71 6 works work NOUN bracis-28371 71 7 by by ADP bracis-28371 71 8 evaluating evaluate VERB bracis-28371 71 9 six six NUM bracis-28371 71 10 algorithms algorithm NOUN bracis-28371 71 11 in in ADP bracis-28371 71 12 new new ADJ bracis-28371 71 13 datasets dataset NOUN bracis-28371 71 14 from from ADP bracis-28371 71 15 two two NUM bracis-28371 71 16 hospitals hospital NOUN bracis-28371 71 17 . . PUNCT bracis-28371 72 1 in in ADP bracis-28371 72 2 this this DET bracis-28371 72 3 work work NOUN bracis-28371 72 4 , , PUNCT bracis-28371 72 5 in in ADP bracis-28371 72 6 addition addition NOUN bracis-28371 72 7 to to ADP bracis-28371 72 8 applying apply VERB bracis-28371 72 9 the the DET bracis-28371 72 10 explainability explainability NOUN bracis-28371 72 11 approach approach NOUN bracis-28371 72 12 , , PUNCT bracis-28371 72 13 we we PRON bracis-28371 72 14 developed develop VERB bracis-28371 72 15 a a DET bracis-28371 72 16 visualization visualization NOUN bracis-28371 72 17 technique technique NOUN bracis-28371 72 18 that that PRON bracis-28371 72 19 allows allow VERB bracis-28371 72 20 drawing draw VERB bracis-28371 72 21 insights insight NOUN bracis-28371 72 22 into into ADP bracis-28371 72 23 the the DET bracis-28371 72 24 relationship relationship NOUN bracis-28371 72 25 between between ADP bracis-28371 72 26 the the DET bracis-28371 72 27 features feature NOUN bracis-28371 72 28 taken take VERB bracis-28371 72 29 into into ADP bracis-28371 72 30 account account NOUN bracis-28371 72 31 by by ADP bracis-28371 72 32 the the DET bracis-28371 72 33 classifiers classifier NOUN bracis-28371 72 34 and and CCONJ bracis-28371 72 35 their their PRON bracis-28371 72 36 correct correct ADJ bracis-28371 72 37 and and CCONJ bracis-28371 72 38 incorrect incorrect ADJ bracis-28371 72 39 predictions prediction NOUN bracis-28371 72 40 . . PUNCT bracis-28371 73 1 3 3 NUM bracis-28371 73 2 materials material NOUN bracis-28371 73 3 and and CCONJ bracis-28371 73 4     SPACE bracis-28371 73 5 methods method NOUN bracis-28371 73 6 this this DET bracis-28371 73 7 work work NOUN bracis-28371 73 8 aims aim VERB bracis-28371 73 9 to to PART bracis-28371 73 10 predict predict VERB bracis-28371 73 11 the the DET bracis-28371 73 12 risk risk NOUN bracis-28371 73 13 of of ADP bracis-28371 73 14 covid-19 covid-19 PROPN bracis-28371 73 15 mortality mortality NOUN bracis-28371 73 16 during during ADP bracis-28371 73 17 patient patient ADJ bracis-28371 73 18 hospitalization hospitalization NOUN bracis-28371 73 19 based base VERB bracis-28371 73 20 on on ADP bracis-28371 73 21 information information NOUN bracis-28371 73 22 available available ADJ bracis-28371 73 23 at at ADP bracis-28371 73 24 the the DET bracis-28371 73 25 time time NOUN bracis-28371 73 26 of of ADP bracis-28371 73 27 admission admission NOUN bracis-28371 73 28 . . PUNCT bracis-28371 74 1 we we PRON bracis-28371 74 2 approach approach VERB bracis-28371 74 3 this this DET bracis-28371 74 4 task task NOUN bracis-28371 74 5 through through ADP bracis-28371 74 6 the the DET bracis-28371 74 7 use use NOUN bracis-28371 74 8 of of ADP bracis-28371 74 9 ml ml NOUN bracis-28371 74 10 classification classification NOUN bracis-28371 74 11 algorithms algorithm NOUN bracis-28371 74 12 . . PUNCT bracis-28371 75 1 our our PRON bracis-28371 75 2 method method NOUN bracis-28371 75 3 comprises comprise VERB bracis-28371 75 4 five five NUM bracis-28371 75 5 steps step NOUN bracis-28371 75 6 , , PUNCT bracis-28371 75 7 depicted depict VERB bracis-28371 75 8 in in ADP bracis-28371 75 9 fig fig NOUN bracis-28371 75 10 . . PUNCT bracis-28371 75 11     SPACE bracis-28371 76 1 1 1 X bracis-28371 76 2 . . PUNCT bracis-28371 77 1 our our PRON bracis-28371 77 2 input input NOUN bracis-28371 77 3 dataset dataset NOUN bracis-28371 77 4 goes go VERB bracis-28371 77 5 through through ADP bracis-28371 77 6 preprocessing preprocessing NOUN bracis-28371 77 7 and and CCONJ bracis-28371 77 8 cleaning cleaning NOUN bracis-28371 77 9 , , PUNCT bracis-28371 77 10 feature feature NOUN bracis-28371 77 11 selection selection NOUN bracis-28371 77 12 , , PUNCT bracis-28371 77 13 then then ADV bracis-28371 77 14 classification classification NOUN bracis-28371 77 15 . . PUNCT bracis-28371 78 1 the the DET bracis-28371 78 2 results result NOUN bracis-28371 78 3 of of ADP bracis-28371 78 4 the the DET bracis-28371 78 5 classifiers classifier NOUN bracis-28371 78 6 are be AUX bracis-28371 78 7 evaluated evaluate VERB bracis-28371 78 8 and and CCONJ bracis-28371 78 9 submitted submit VERB bracis-28371 78 10 to to ADP bracis-28371 78 11 the the DET bracis-28371 78 12 explainability explainability NOUN bracis-28371 78 13 approach approach NOUN bracis-28371 78 14 . . PUNCT bracis-28371 79 1 the the DET bracis-28371 79 2 next next ADJ bracis-28371 79 3 sections section NOUN bracis-28371 79 4 describe describe VERB bracis-28371 79 5 our our PRON bracis-28371 79 6 data datum NOUN bracis-28371 79 7 and and CCONJ bracis-28371 79 8 each each DET bracis-28371 79 9 step step NOUN bracis-28371 79 10 in in ADP bracis-28371 79 11 greater great ADJ bracis-28371 79 12 detail detail NOUN bracis-28371 79 13 . . PUNCT bracis-28371 80 1 fig fig NOUN bracis-28371 80 2 . . PUNCT bracis-28371 81 1 1 1 X bracis-28371 81 2 . . X bracis-28371 81 3 pipeline pipeline NOUN bracis-28371 81 4 employed employ VERB bracis-28371 81 5 in in ADP bracis-28371 81 6 this this DET bracis-28371 81 7 work work NOUN bracis-28371 81 8 . . PUNCT bracis-28371 82 1 full full ADJ bracis-28371 82 2 size size NOUN bracis-28371 82 3 image image NOUN bracis-28371 82 4 3.1 3.1 NUM bracis-28371 82 5 patient patient NOUN bracis-28371 82 6 records record NOUN bracis-28371 82 7 and and CCONJ bracis-28371 82 8     SPACE bracis-28371 82 9 preprocessing preprocesse VERB bracis-28371 82 10 the the DET bracis-28371 82 11 study study NOUN bracis-28371 82 12 involved involve VERB bracis-28371 82 13 patients patient NOUN bracis-28371 82 14 hospitalized hospitalize VERB bracis-28371 82 15 in in ADP bracis-28371 82 16 two two NUM bracis-28371 82 17 institutions institution NOUN bracis-28371 82 18 : : PUNCT bracis-28371 82 19 hmv hmv PROPN bracis-28371 82 20 and and CCONJ bracis-28371 82 21 hcpa hcpa VERB bracis-28371 82 22 . . PUNCT bracis-28371 83 1 hmv hmv PROPN bracis-28371 83 2 is be AUX bracis-28371 83 3 a a DET bracis-28371 83 4 private private ADJ bracis-28371 83 5 hospital hospital NOUN bracis-28371 83 6 . . PUNCT bracis-28371 84 1 hcpa hcpa PROPN bracis-28371 84 2 is be AUX bracis-28371 84 3 a a DET bracis-28371 84 4 public public ADJ bracis-28371 84 5 tertiary tertiary ADJ bracis-28371 84 6 care care NOUN bracis-28371 84 7 teaching teaching NOUN bracis-28371 84 8 hospital hospital NOUN bracis-28371 84 9 academically academically ADV bracis-28371 84 10 linked link VERB bracis-28371 84 11 to to ADP bracis-28371 84 12 a a DET bracis-28371 84 13 university university NOUN bracis-28371 84 14 . . PUNCT bracis-28371 85 1 the the DET bracis-28371 85 2 project project NOUN bracis-28371 85 3 was be AUX bracis-28371 85 4 approved approve VERB bracis-28371 85 5 by by ADP bracis-28371 85 6 the the DET bracis-28371 85 7 institutional institutional ADJ bracis-28371 85 8 research research PROPN bracis-28371 85 9 ethics ethic NOUN bracis-28371 85 10 committee committee PROPN bracis-28371 85 11 of of ADP bracis-28371 85 12 both both PRON bracis-28371 85 13 institutionsfootnote institutionsfootnote VERB bracis-28371 85 14 1 1 NUM bracis-28371 85 15 . . PUNCT bracis-28371 85 16 informed informed ADJ bracis-28371 85 17 consent consent NOUN bracis-28371 85 18 was be AUX bracis-28371 85 19 waived waive VERB bracis-28371 85 20 due due ADP bracis-28371 85 21 to to ADP bracis-28371 85 22 the the DET bracis-28371 85 23 retrospective retrospective ADJ bracis-28371 85 24 nature nature NOUN bracis-28371 85 25 of of ADP bracis-28371 85 26 the the DET bracis-28371 85 27 study study NOUN bracis-28371 85 28 . . PUNCT bracis-28371 86 1 the the DET bracis-28371 86 2 collected collect VERB bracis-28371 86 3 data datum NOUN bracis-28371 86 4 did do AUX bracis-28371 86 5 not not PART bracis-28371 86 6 include include VERB bracis-28371 86 7 any any DET bracis-28371 86 8 identifiable identifiable ADJ bracis-28371 86 9 information information NOUN bracis-28371 86 10 to to PART bracis-28371 86 11 ensure ensure VERB bracis-28371 86 12 patient patient ADJ bracis-28371 86 13 privacy privacy NOUN bracis-28371 86 14 . . PUNCT bracis-28371 87 1 patients patient NOUN bracis-28371 87 2 included include VERB bracis-28371 87 3 in in ADP bracis-28371 87 4 the the DET bracis-28371 87 5 study study NOUN bracis-28371 87 6 are be AUX bracis-28371 87 7 adults adult NOUN bracis-28371 87 8 ( ( PUNCT bracis-28371 87 9 age age NOUN bracis-28371 87 10     SPACE bracis-28371 87 11 \(\ge \(\ge PROPN bracis-28371 87 12 \ \ PROPN bracis-28371 87 13 ) ) PUNCT bracis-28371 87 14     SPACE bracis-28371 87 15 18 18 NUM bracis-28371 87 16 ) ) PUNCT bracis-28371 87 17 diagnosed diagnose VERB bracis-28371 87 18 with with ADP bracis-28371 87 19 covid-19 covid-19 PROPN bracis-28371 87 20 using use VERB bracis-28371 87 21 the the DET bracis-28371 87 22 reverse reverse ADJ bracis-28371 87 23 transcriptase transcriptase NOUN bracis-28371 87 24 polymerase polymerase NOUN bracis-28371 87 25 chain chain NOUN bracis-28371 87 26 reaction reaction NOUN bracis-28371 87 27 ( ( PUNCT bracis-28371 87 28 rt rt ADJ bracis-28371 87 29 - - PUNCT bracis-28371 87 30 pcr pcr NOUN bracis-28371 87 31 ) ) PUNCT bracis-28371 87 32 test test NOUN bracis-28371 87 33 . . PUNCT bracis-28371 88 1 the the DET bracis-28371 88 2 features feature NOUN bracis-28371 88 3 include include VERB bracis-28371 88 4 clinical clinical ADJ bracis-28371 88 5 and and CCONJ bracis-28371 88 6 demographic demographic ADJ bracis-28371 88 7 data datum NOUN bracis-28371 88 8 : : PUNCT bracis-28371 88 9 age age NOUN bracis-28371 88 10 , , PUNCT bracis-28371 88 11 sex sex NOUN bracis-28371 88 12 , , PUNCT bracis-28371 88 13 ethnicity ethnicity NOUN bracis-28371 88 14 , , PUNCT bracis-28371 88 15 comorbidities comorbiditie NOUN bracis-28371 88 16 ( ( PUNCT bracis-28371 88 17 hypertension hypertension NOUN bracis-28371 88 18 , , PUNCT bracis-28371 88 19 obesity obesity NOUN bracis-28371 88 20 , , PUNCT bracis-28371 88 21 diabetes diabetes NOUN bracis-28371 88 22 , , PUNCT bracis-28371 88 23 etc etc X bracis-28371 88 24 . . X bracis-28371 88 25 ) ) PUNCT bracis-28371 88 26 , , PUNCT bracis-28371 88 27 lab lab NOUN bracis-28371 88 28 tests test NOUN bracis-28371 88 29 ( ( PUNCT bracis-28371 88 30 arterial arterial ADJ bracis-28371 88 31 blood blood NOUN bracis-28371 88 32 gas gas NOUN bracis-28371 88 33 , , PUNCT bracis-28371 88 34 lactate lactate NOUN bracis-28371 88 35 , , PUNCT bracis-28371 88 36 blood blood NOUN bracis-28371 88 37 count count NOUN bracis-28371 88 38 , , PUNCT bracis-28371 88 39 platelets platelet NOUN bracis-28371 88 40 , , PUNCT bracis-28371 88 41 c c NOUN bracis-28371 88 42 - - PUNCT bracis-28371 88 43 reactive reactive ADJ bracis-28371 88 44 protein protein NOUN bracis-28371 88 45 , , PUNCT bracis-28371 88 46 d d NOUN bracis-28371 88 47 - - PUNCT bracis-28371 88 48 dimer dimer ADJ bracis-28371 88 49 , , PUNCT bracis-28371 88 50 creatine creatine PROPN bracis-28371 88 51 kinase kinase PROPN bracis-28371 88 52 ( ( PUNCT bracis-28371 88 53 ck ck PROPN bracis-28371 88 54 ) ) PUNCT bracis-28371 88 55 , , PUNCT bracis-28371 88 56 creatinine creatinine NOUN bracis-28371 88 57 , , PUNCT bracis-28371 88 58 urea urea NOUN bracis-28371 88 59 , , PUNCT bracis-28371 88 60 sodium sodium NOUN bracis-28371 88 61 , , PUNCT bracis-28371 88 62 potassium potassium NOUN bracis-28371 88 63 , , PUNCT bracis-28371 88 64 bilirubin bilirubin NOUN bracis-28371 88 65 ) ) PUNCT bracis-28371 88 66 , , PUNCT bracis-28371 88 67 need need VERB bracis-28371 88 68 for for ADP bracis-28371 88 69 icu icu PROPN bracis-28371 88 70 admission admission NOUN bracis-28371 88 71 , , PUNCT bracis-28371 88 72 type type NOUN bracis-28371 88 73 of of ADP bracis-28371 88 74 ventilatory ventilatory ADJ bracis-28371 88 75 support support NOUN bracis-28371 88 76 during during ADP bracis-28371 88 77 hospitalization hospitalization NOUN bracis-28371 88 78 and and CCONJ bracis-28371 88 79 icu icu PROPN bracis-28371 88 80 period period NOUN bracis-28371 88 81 ( ( PUNCT bracis-28371 88 82 o2 o2 PROPN bracis-28371 88 83 catheter catheter NOUN bracis-28371 88 84 , , PUNCT bracis-28371 88 85 non non ADJ bracis-28371 88 86 - - ADJ bracis-28371 88 87 invasive invasive ADJ bracis-28371 88 88 ventilation ventilation NOUN bracis-28371 88 89 , , PUNCT bracis-28371 88 90 high high ADJ bracis-28371 88 91 - - PUNCT bracis-28371 88 92 flow flow NOUN bracis-28371 88 93 nasal nasal NOUN bracis-28371 88 94 cannula cannula NOUN bracis-28371 88 95 , , PUNCT bracis-28371 88 96 invasive invasive ADJ bracis-28371 88 97 mechanical mechanical ADJ bracis-28371 88 98 ventilation ventilation NOUN bracis-28371 88 99 ) ) PUNCT bracis-28371 88 100 , , PUNCT bracis-28371 88 101 start start VERB bracis-28371 88 102 and and CCONJ bracis-28371 88 103 end end VERB bracis-28371 88 104 date date NOUN bracis-28371 88 105 of of ADP bracis-28371 88 106 hospitalization hospitalization NOUN bracis-28371 88 107 , , PUNCT bracis-28371 88 108 as as ADV bracis-28371 88 109 well well ADV bracis-28371 88 110 as as ADP bracis-28371 88 111 hospitalization hospitalization NOUN bracis-28371 88 112 outcome outcome NOUN bracis-28371 88 113 . . PUNCT bracis-28371 89 1 we we PRON bracis-28371 89 2 performed perform VERB bracis-28371 89 3 the the DET bracis-28371 89 4 shapiro shapiro PROPN bracis-28371 89 5 - - PUNCT bracis-28371 89 6 wilk wilk NOUN bracis-28371 89 7 test test NOUN bracis-28371 89 8 on on ADP bracis-28371 89 9 continuous continuous ADJ bracis-28371 89 10 features feature NOUN bracis-28371 89 11 . . PUNCT bracis-28371 90 1 the the DET bracis-28371 90 2 results result NOUN bracis-28371 90 3 showed show VERB bracis-28371 90 4 that that SCONJ bracis-28371 90 5 our our PRON bracis-28371 90 6 features feature NOUN bracis-28371 90 7 are be AUX bracis-28371 90 8 not not PART bracis-28371 90 9 normally normally ADV bracis-28371 90 10 distributed distribute VERB bracis-28371 90 11 ( ( PUNCT bracis-28371 90 12 \(\alpha \(\alpha NOUN bracis-28371 90 13 = = SYM bracis-28371 90 14 0.01\ 0.01\ NUM bracis-28371 90 15 ) ) PUNCT bracis-28371 90 16 ) ) PUNCT bracis-28371 90 17 . . PUNCT bracis-28371 91 1 patients patient NOUN bracis-28371 91 2 without without ADP bracis-28371 91 3 information information NOUN bracis-28371 91 4 on on ADP bracis-28371 91 5 the the DET bracis-28371 91 6 outcome outcome NOUN bracis-28371 91 7 of of ADP bracis-28371 91 8 hospitalization hospitalization NOUN bracis-28371 91 9 ( ( PUNCT bracis-28371 91 10 dependent dependent ADJ bracis-28371 91 11 variable variable NOUN bracis-28371 91 12 ) ) PUNCT bracis-28371 91 13 were be AUX bracis-28371 91 14 not not PART bracis-28371 91 15 considered consider VERB bracis-28371 91 16 in in ADP bracis-28371 91 17 this this DET bracis-28371 91 18 study study NOUN bracis-28371 91 19 . . PUNCT bracis-28371 92 1 at at ADP bracis-28371 92 2 the the DET bracis-28371 92 3 hcpa hcpa ADJ bracis-28371 92 4 hospital hospital NOUN bracis-28371 92 5 , , PUNCT bracis-28371 92 6 318 318 NUM bracis-28371 92 7 patients patient NOUN bracis-28371 92 8 had have VERB bracis-28371 92 9 no no DET bracis-28371 92 10 record record NOUN bracis-28371 92 11 of of ADP bracis-28371 92 12 outcome outcome NOUN bracis-28371 92 13 due due ADJ bracis-28371 92 14 to to PART bracis-28371 92 15 transfer transfer VERB bracis-28371 92 16 to to ADP bracis-28371 92 17 other other ADJ bracis-28371 92 18 hospitals hospital NOUN bracis-28371 92 19 and and CCONJ bracis-28371 92 20 6 6 NUM bracis-28371 92 21 due due ADP bracis-28371 92 22 to to ADP bracis-28371 92 23 withdrawal withdrawal NOUN bracis-28371 92 24 or or CCONJ bracis-28371 92 25 evasion evasion NOUN bracis-28371 92 26 of of ADP bracis-28371 92 27 treatment treatment NOUN bracis-28371 92 28 . . PUNCT bracis-28371 93 1 for for ADP bracis-28371 93 2 the the DET bracis-28371 93 3 hmv hmv PROPN bracis-28371 93 4 hospital hospital NOUN bracis-28371 93 5 , , PUNCT bracis-28371 93 6 patients patient NOUN bracis-28371 93 7 without without ADP bracis-28371 93 8 outcome outcome NOUN bracis-28371 93 9 information information NOUN bracis-28371 93 10 were be AUX bracis-28371 93 11 disregarded disregard VERB bracis-28371 93 12 during during ADP bracis-28371 93 13 data data NOUN bracis-28371 93 14 collection collection NOUN bracis-28371 93 15 . . PUNCT bracis-28371 94 1 variables variable NOUN bracis-28371 94 2 with with ADP bracis-28371 94 3 missing miss VERB bracis-28371 94 4 values value NOUN bracis-28371 94 5 greater great ADJ bracis-28371 94 6 than than ADP bracis-28371 94 7 40 40 NUM bracis-28371 94 8 % % NOUN bracis-28371 94 9 were be AUX bracis-28371 94 10 also also ADV bracis-28371 94 11 not not PART bracis-28371 94 12 considered consider VERB bracis-28371 94 13 . . PUNCT bracis-28371 95 1 in in ADP bracis-28371 95 2 the the DET bracis-28371 95 3 hmv hmv PROPN bracis-28371 95 4 dataset dataset VERB bracis-28371 95 5 the the DET bracis-28371 95 6 variable variable ADJ bracis-28371 95 7 brain brain NOUN bracis-28371 95 8 natriuretic natriuretic ADJ bracis-28371 95 9 peptide peptide NOUN bracis-28371 95 10 was be AUX bracis-28371 95 11 discarded discard VERB bracis-28371 95 12 , , PUNCT bracis-28371 95 13 as as ADV bracis-28371 95 14 well well ADV bracis-28371 95 15 as as ADP bracis-28371 95 16 the the DET bracis-28371 95 17 variables variable NOUN bracis-28371 95 18 weight weight NOUN bracis-28371 95 19 , , PUNCT bracis-28371 95 20 height height NOUN bracis-28371 95 21 and and CCONJ bracis-28371 95 22 blood blood NOUN bracis-28371 95 23 type type NOUN bracis-28371 95 24 for for ADP bracis-28371 95 25 hcpa hcpa NOUN bracis-28371 95 26 . . PUNCT bracis-28371 96 1 the the DET bracis-28371 96 2 variables variable NOUN bracis-28371 96 3 troponin troponin ADJ bracis-28371 96 4 from from ADP bracis-28371 96 5 hmv hmv PROPN bracis-28371 96 6 and and CCONJ bracis-28371 96 7 fibrinogen fibrinogen NOUN bracis-28371 96 8 from from ADP bracis-28371 96 9 hcpa hcpa NOUN bracis-28371 96 10 were be AUX bracis-28371 96 11 the the DET bracis-28371 96 12 only only ADJ bracis-28371 96 13 exceptions exception NOUN bracis-28371 96 14 to to ADP bracis-28371 96 15 this this DET bracis-28371 96 16 rule rule NOUN bracis-28371 96 17 , , PUNCT bracis-28371 96 18 because because SCONJ bracis-28371 96 19 although although SCONJ bracis-28371 96 20 they they PRON bracis-28371 96 21 had have VERB bracis-28371 96 22 a a DET bracis-28371 96 23 missing missing ADJ bracis-28371 96 24 value value NOUN bracis-28371 96 25 rate rate NOUN bracis-28371 96 26 of of ADP bracis-28371 96 27 44.67 44.67 NUM bracis-28371 96 28 % % NOUN bracis-28371 96 29 and and CCONJ bracis-28371 96 30 41.78 41.78 NUM bracis-28371 96 31 % % NOUN bracis-28371 96 32 , , PUNCT bracis-28371 96 33 respectively respectively ADV bracis-28371 96 34 , , PUNCT bracis-28371 96 35 there there PRON bracis-28371 96 36 is be VERB bracis-28371 96 37 previous previous ADJ bracis-28371 96 38 evidence evidence NOUN bracis-28371 96 39 regarding regard VERB bracis-28371 96 40 their their PRON bracis-28371 96 41 importance importance NOUN bracis-28371 96 42 as as ADP bracis-28371 96 43 indicators indicator NOUN bracis-28371 96 44 of of ADP bracis-28371 96 45 severity severity NOUN bracis-28371 96 46 for for ADP bracis-28371 96 47 covid-19 covid-19 PROPN bracis-28371 96 48 . . PROPN bracis-28371 96 49 in in ADP bracis-28371 96 50 hmv hmv PROPN bracis-28371 96 51 , , PUNCT bracis-28371 96 52 23 23 NUM bracis-28371 96 53 patients patient NOUN bracis-28371 96 54 ( ( PUNCT bracis-28371 96 55 5 5 NUM bracis-28371 96 56 deaths death NOUN bracis-28371 96 57 and and CCONJ bracis-28371 96 58 18 18 NUM bracis-28371 96 59 discharges discharge NOUN bracis-28371 96 60 ) ) PUNCT bracis-28371 96 61 were be AUX bracis-28371 96 62 disregarded disregard VERB bracis-28371 96 63 because because SCONJ bracis-28371 96 64 they they PRON bracis-28371 96 65 had have VERB bracis-28371 96 66 a a DET bracis-28371 96 67 missing missing ADJ bracis-28371 96 68 value value NOUN bracis-28371 96 69 index index NOUN bracis-28371 96 70 greater great ADJ bracis-28371 96 71 than than ADP bracis-28371 96 72 30 30 NUM bracis-28371 96 73 % % NOUN bracis-28371 96 74 , , PUNCT bracis-28371 96 75 while while SCONJ bracis-28371 96 76 in in ADP bracis-28371 96 77 hcpa hcpa NOUN bracis-28371 96 78 , , PUNCT bracis-28371 96 79 258 258 NUM bracis-28371 96 80 ( ( PUNCT bracis-28371 96 81 93 93 NUM bracis-28371 96 82 deaths death NOUN bracis-28371 96 83 and and CCONJ bracis-28371 96 84 165 165 NUM bracis-28371 96 85 discharges discharge NOUN bracis-28371 96 86 ) ) PUNCT bracis-28371 96 87 were be AUX bracis-28371 96 88 not not PART bracis-28371 96 89 considered consider VERB bracis-28371 96 90 . . PUNCT bracis-28371 97 1 in in ADP bracis-28371 97 2 hmv hmv PROPN bracis-28371 97 3 and and CCONJ bracis-28371 97 4 hcpa hcpa VERB bracis-28371 97 5 , , PUNCT bracis-28371 97 6 7 7 NUM bracis-28371 97 7 and and CCONJ bracis-28371 97 8 19 19 NUM bracis-28371 97 9 patients patient NOUN bracis-28371 97 10 , , PUNCT bracis-28371 97 11 respectively respectively ADV bracis-28371 97 12 , , PUNCT bracis-28371 97 13 had have VERB bracis-28371 97 14 two two NUM bracis-28371 97 15 hospital hospital NOUN bracis-28371 97 16 admissions admission NOUN bracis-28371 97 17 ( ( PUNCT bracis-28371 97 18 readmissions readmission NOUN bracis-28371 97 19 ) ) PUNCT bracis-28371 97 20 . . PUNCT bracis-28371 98 1 among among ADP bracis-28371 98 2 these these PRON bracis-28371 98 3 , , PUNCT bracis-28371 98 4 7 7 NUM bracis-28371 98 5 patients patient NOUN bracis-28371 98 6 from from ADP bracis-28371 98 7 hmv hmv PROPN bracis-28371 98 8 and and CCONJ bracis-28371 98 9 13 13 NUM bracis-28371 98 10 patients patient NOUN bracis-28371 98 11 from from ADP bracis-28371 98 12 hcpa hcpa NOUN bracis-28371 98 13 were be AUX bracis-28371 98 14 readmitted readmitte VERB bracis-28371 98 15 within within ADP bracis-28371 98 16 30 30 NUM bracis-28371 98 17 days day NOUN bracis-28371 98 18 of of ADP bracis-28371 98 19 the the DET bracis-28371 98 20 discharge discharge NOUN bracis-28371 98 21 from from ADP bracis-28371 98 22 the the DET bracis-28371 98 23 index index NOUN bracis-28371 98 24 hospitalization hospitalization NOUN bracis-28371 98 25 . . PUNCT bracis-28371 99 1 for for ADP bracis-28371 99 2 these these DET bracis-28371 99 3 patients patient NOUN bracis-28371 99 4 , , PUNCT bracis-28371 99 5 we we PRON bracis-28371 99 6 used use VERB bracis-28371 99 7 the the DET bracis-28371 99 8 admission admission NOUN bracis-28371 99 9 records record NOUN bracis-28371 99 10 collected collect VERB bracis-28371 99 11 at at ADP bracis-28371 99 12 the the DET bracis-28371 99 13 index index NOUN bracis-28371 99 14 hospitalization hospitalization NOUN bracis-28371 99 15 with with ADP bracis-28371 99 16 the the DET bracis-28371 99 17 readmission readmission NOUN bracis-28371 99 18 outcome outcome NOUN bracis-28371 99 19 as as ADP bracis-28371 99 20 the the DET bracis-28371 99 21 label label NOUN bracis-28371 99 22 ( ( PUNCT bracis-28371 99 23 patients patient NOUN bracis-28371 99 24 who who PRON bracis-28371 99 25 were be AUX bracis-28371 99 26 discharged discharge VERB bracis-28371 99 27 , , PUNCT bracis-28371 99 28 returned return VERB bracis-28371 99 29 within within ADP bracis-28371 99 30 30 30 NUM bracis-28371 99 31 days day NOUN bracis-28371 99 32 , , PUNCT bracis-28371 99 33 and and CCONJ bracis-28371 99 34 died die VERB bracis-28371 99 35 , , PUNCT bracis-28371 99 36 were be AUX bracis-28371 99 37 considered consider VERB bracis-28371 99 38 deaths death NOUN bracis-28371 99 39 ) ) PUNCT bracis-28371 99 40 . . PUNCT bracis-28371 100 1 for for ADP bracis-28371 100 2 the the DET bracis-28371 100 3 other other ADJ bracis-28371 100 4 6 6 NUM bracis-28371 100 5 patients patient NOUN bracis-28371 100 6 from from ADP bracis-28371 100 7 hcpa hcpa NOUN bracis-28371 100 8 who who PRON bracis-28371 100 9 were be AUX bracis-28371 100 10 readmitted readmitte VERB bracis-28371 100 11 after after ADP bracis-28371 100 12 30 30 NUM bracis-28371 100 13 days day NOUN bracis-28371 100 14 , , PUNCT bracis-28371 100 15 we we PRON bracis-28371 100 16 used use VERB bracis-28371 100 17 only only ADV bracis-28371 100 18 the the DET bracis-28371 100 19 record record NOUN bracis-28371 100 20 of of ADP bracis-28371 100 21 the the DET bracis-28371 100 22 index index NOUN bracis-28371 100 23 hospitalization hospitalization NOUN bracis-28371 100 24 . . PUNCT bracis-28371 101 1 hmv hmv PROPN bracis-28371 101 2 datasets datasets PROPN bracis-28371 101 3 . . PUNCT bracis-28371 102 1 contain contain VERB bracis-28371 102 2 the the DET bracis-28371 102 3 anonymized anonymize VERB bracis-28371 102 4 records record NOUN bracis-28371 102 5 of of ADP bracis-28371 102 6 patients patient NOUN bracis-28371 102 7 confirmed confirm VERB bracis-28371 102 8 for for ADP bracis-28371 102 9 covid-19 covid-19 PROPN bracis-28371 102 10 admitted admit VERB bracis-28371 102 11 at at ADP bracis-28371 102 12 hmv hmv PROPN bracis-28371 102 13 between between ADP bracis-28371 102 14 march march PROPN bracis-28371 102 15 2020 2020 NUM bracis-28371 102 16 and and CCONJ bracis-28371 102 17 june june PROPN bracis-28371 102 18 2021 2021 NUM bracis-28371 102 19 . . PUNCT bracis-28371 103 1 there there PRON bracis-28371 103 2 are be VERB bracis-28371 103 3 1,526 1,526 NUM bracis-28371 103 4 records record NOUN bracis-28371 103 5 , , PUNCT bracis-28371 103 6 among among ADP bracis-28371 103 7 which which PRON bracis-28371 103 8 58.78 58.78 NUM bracis-28371 103 9 % % NOUN bracis-28371 103 10 refer refer VERB bracis-28371 103 11 to to ADP bracis-28371 103 12 male male ADJ bracis-28371 103 13 patients patient NOUN bracis-28371 103 14 . . PUNCT bracis-28371 104 1 the the DET bracis-28371 104 2 dataset dataset NOUN bracis-28371 104 3 covers cover VERB bracis-28371 104 4 both both DET bracis-28371 104 5 ward ward NOUN bracis-28371 104 6 patients patient NOUN bracis-28371 104 7 and and CCONJ bracis-28371 104 8 patients patient NOUN bracis-28371 104 9 admitted admit VERB bracis-28371 104 10 to to ADP bracis-28371 104 11 the the DET bracis-28371 104 12 icu icu PROPN bracis-28371 104 13 . . PUNCT bracis-28371 105 1 the the DET bracis-28371 105 2 data data NOUN bracis-28371 105 3 collection collection NOUN bracis-28371 105 4 process process NOUN bracis-28371 105 5 was be AUX bracis-28371 105 6 conducted conduct VERB bracis-28371 105 7 by by ADP bracis-28371 105 8 health health NOUN bracis-28371 105 9 professionals professional NOUN bracis-28371 105 10 , , PUNCT bracis-28371 105 11 who who PRON bracis-28371 105 12 manually manually ADV bracis-28371 105 13 transcribed transcribe VERB bracis-28371 105 14 the the DET bracis-28371 105 15 information information NOUN bracis-28371 105 16 contained contain VERB bracis-28371 105 17 in in ADP bracis-28371 105 18 the the DET bracis-28371 105 19 patients patient NOUN bracis-28371 105 20 ’ ' PUNCT bracis-28371 105 21 medical medical ADJ bracis-28371 105 22 records record NOUN bracis-28371 105 23 into into ADP bracis-28371 105 24 a a DET bracis-28371 105 25 form form NOUN bracis-28371 105 26 containing contain VERB bracis-28371 105 27 the the DET bracis-28371 105 28 features feature NOUN bracis-28371 105 29 previously previously ADV bracis-28371 105 30 established establish VERB bracis-28371 105 31 for for ADP bracis-28371 105 32 the the DET bracis-28371 105 33 study study NOUN bracis-28371 105 34 ( ( PUNCT bracis-28371 105 35 clinical clinical ADJ bracis-28371 105 36 information information NOUN bracis-28371 105 37 available available ADJ bracis-28371 105 38 at at ADP bracis-28371 105 39 hospital hospital NOUN bracis-28371 105 40 admission admission NOUN bracis-28371 105 41 ) ) PUNCT bracis-28371 105 42 . . PUNCT bracis-28371 106 1 patients patient NOUN bracis-28371 106 2 without without ADP bracis-28371 106 3 outcome outcome NOUN bracis-28371 106 4 information information NOUN bracis-28371 106 5 were be AUX bracis-28371 106 6 not not PART bracis-28371 106 7 considered consider VERB bracis-28371 106 8 during during ADP bracis-28371 106 9 the the DET bracis-28371 106 10 data data NOUN bracis-28371 106 11 transcription transcription NOUN bracis-28371 106 12 process process NOUN bracis-28371 106 13 . . PUNCT bracis-28371 107 1 thus thus ADV bracis-28371 107 2 , , PUNCT bracis-28371 107 3 our our PRON bracis-28371 107 4 study study NOUN bracis-28371 107 5 did do AUX bracis-28371 107 6 not not PART bracis-28371 107 7 include include VERB bracis-28371 107 8 hospitalization hospitalization NOUN bracis-28371 107 9 records record NOUN bracis-28371 107 10 in in ADP bracis-28371 107 11 which which PRON bracis-28371 107 12 patients patient NOUN bracis-28371 107 13 were be AUX bracis-28371 107 14 transferred transfer VERB bracis-28371 107 15 to to ADP bracis-28371 107 16 other other ADJ bracis-28371 107 17 hospitals hospital NOUN bracis-28371 107 18 and/or and/or CCONJ bracis-28371 107 19 dropout dropout NOUN bracis-28371 107 20 cases case NOUN bracis-28371 107 21 . . PUNCT bracis-28371 108 1 automatic automatic ADJ bracis-28371 108 2 verification verification NOUN bracis-28371 108 3 processes process NOUN bracis-28371 108 4 were be AUX bracis-28371 108 5 applied apply VERB bracis-28371 108 6 to to PART bracis-28371 108 7 identify identify VERB bracis-28371 108 8 inconsistencies inconsistency NOUN bracis-28371 108 9 . . PUNCT bracis-28371 109 1 we we PRON bracis-28371 109 2 developed develop VERB bracis-28371 109 3 classifiers classifier NOUN bracis-28371 109 4 considering consider VERB bracis-28371 109 5 two two NUM bracis-28371 109 6 scenarios scenario NOUN bracis-28371 109 7 . . PUNCT bracis-28371 110 1 in in ADP bracis-28371 110 2 the the DET bracis-28371 110 3 first first ADJ bracis-28371 110 4 scenario scenario NOUN bracis-28371 110 5 , , PUNCT bracis-28371 110 6 data datum NOUN bracis-28371 110 7 of of ADP bracis-28371 110 8 all all DET bracis-28371 110 9 patients patient NOUN bracis-28371 110 10 are be AUX bracis-28371 110 11 considered consider VERB bracis-28371 110 12 , , PUNCT bracis-28371 110 13 including include VERB bracis-28371 110 14 ward ward NOUN bracis-28371 110 15 and and CCONJ bracis-28371 110 16 icu icu PROPN bracis-28371 110 17 patients patient NOUN bracis-28371 110 18 ( ( PUNCT bracis-28371 110 19 \(\text \(\text NOUN bracis-28371 110 20 { { PUNCT bracis-28371 110 21 hmv}_{all}\ hmv}_{all}\ PROPN bracis-28371 110 22 ) ) PUNCT bracis-28371 110 23 ) ) PUNCT bracis-28371 110 24 . . PUNCT bracis-28371 111 1 in in ADP bracis-28371 111 2 the the DET bracis-28371 111 3 second second ADJ bracis-28371 111 4 scenario scenario NOUN bracis-28371 111 5 , , PUNCT bracis-28371 111 6 only only ADV bracis-28371 111 7 patients patient NOUN bracis-28371 111 8 admitted admit VERB bracis-28371 111 9 to to ADP bracis-28371 111 10 the the DET bracis-28371 111 11 icu icu PROPN bracis-28371 111 12 are be AUX bracis-28371 111 13 considered consider VERB bracis-28371 111 14 ( ( PUNCT bracis-28371 111 15 \(\text \(\text NOUN bracis-28371 111 16 { { PUNCT bracis-28371 111 17 hmv}_{cti}\ hmv}_{cti}\ PROPN bracis-28371 111 18 ) ) PUNCT bracis-28371 111 19 ) ) PUNCT bracis-28371 111 20 . . PUNCT bracis-28371 112 1 in in ADP bracis-28371 112 2 both both DET bracis-28371 112 3 cases case NOUN bracis-28371 112 4 , , PUNCT bracis-28371 112 5 we we PRON bracis-28371 112 6 focus focus VERB bracis-28371 112 7 on on ADP bracis-28371 112 8 information information NOUN bracis-28371 112 9 available available ADJ bracis-28371 112 10 during during ADP bracis-28371 112 11 admission admission NOUN bracis-28371 112 12 . . PUNCT bracis-28371 113 1 \({\textbf \({\textbf PROPN bracis-28371 113 2 { { PUNCT bracis-28371 113 3 hcpa}}_{\boldsymbol{cti}}\ hcpa}}_{\boldsymbol{cti}}\ PROPN bracis-28371 113 4 ) ) PUNCT bracis-28371 113 5 dataset dataset NOUN bracis-28371 113 6 contains contain VERB bracis-28371 113 7 anonymized anonymize VERB bracis-28371 113 8 records record NOUN bracis-28371 113 9 of of ADP bracis-28371 113 10 2,269 2,269 NUM bracis-28371 113 11 patients patient NOUN bracis-28371 113 12 with with ADP bracis-28371 113 13 confirmed confirm VERB bracis-28371 113 14 covid-19 covid-19 PROPN bracis-28371 113 15 admitted admit VERB bracis-28371 113 16 to to ADP bracis-28371 113 17 the the DET bracis-28371 113 18 icu icu PROPN bracis-28371 113 19 of of ADP bracis-28371 113 20 hcpa hcpa NOUN bracis-28371 113 21 between between ADP bracis-28371 113 22 march march PROPN bracis-28371 113 23 2020 2020 NUM bracis-28371 113 24 and and CCONJ bracis-28371 113 25 august august PROPN bracis-28371 113 26 2021 2021 NUM bracis-28371 113 27 . . PUNCT bracis-28371 114 1 male male ADJ bracis-28371 114 2 patients patient NOUN bracis-28371 114 3 represent represent VERB bracis-28371 114 4 54.69 54.69 NUM bracis-28371 114 5 % % NOUN bracis-28371 114 6 of of ADP bracis-28371 114 7 the the DET bracis-28371 114 8 dataset dataset NOUN bracis-28371 114 9 . . PUNCT bracis-28371 115 1 the the DET bracis-28371 115 2 data datum NOUN bracis-28371 115 3 was be AUX bracis-28371 115 4 collected collect VERB bracis-28371 115 5 directly directly ADV bracis-28371 115 6 from from ADP bracis-28371 115 7 the the DET bracis-28371 115 8 hospital hospital NOUN bracis-28371 115 9 database database NOUN bracis-28371 115 10 . . PUNCT bracis-28371 116 1 the the DET bracis-28371 116 2 outcome outcome NOUN bracis-28371 116 3 of of ADP bracis-28371 116 4 interest interest NOUN bracis-28371 116 5 was be AUX bracis-28371 116 6 mortality mortality NOUN bracis-28371 116 7 during during ADP bracis-28371 116 8 hospitalization hospitalization NOUN bracis-28371 116 9 , , PUNCT bracis-28371 116 10 thus thus ADV bracis-28371 116 11 , , PUNCT bracis-28371 116 12 records record NOUN bracis-28371 116 13 were be AUX bracis-28371 116 14 categorized categorize VERB bracis-28371 116 15 as as ADP bracis-28371 116 16 alive alive ADJ bracis-28371 116 17 or or CCONJ bracis-28371 116 18 deceased deceased ADJ bracis-28371 116 19 according accord VERB bracis-28371 116 20 to to ADP bracis-28371 116 21 data datum NOUN bracis-28371 116 22 provided provide VERB bracis-28371 116 23 by by ADP bracis-28371 116 24 the the DET bracis-28371 116 25 hospitals hospital NOUN bracis-28371 116 26 . . PUNCT bracis-28371 117 1 details detail NOUN bracis-28371 117 2 of of ADP bracis-28371 117 3 the the DET bracis-28371 117 4 datasets dataset NOUN bracis-28371 117 5 after after SCONJ bracis-28371 117 6 the the DET bracis-28371 117 7 preprocessing preprocessing NOUN bracis-28371 117 8 step step NOUN bracis-28371 117 9 are be AUX bracis-28371 117 10 shown show VERB bracis-28371 117 11 in in ADP bracis-28371 117 12 table table NOUN bracis-28371 117 13     SPACE bracis-28371 117 14 1 1 NUM bracis-28371 117 15 . . PUNCT bracis-28371 117 16 table table NOUN bracis-28371 117 17 1 1 NUM bracis-28371 117 18 . . PUNCT bracis-28371 118 1 statistics statistic NOUN bracis-28371 118 2 of of ADP bracis-28371 118 3 the the DET bracis-28371 118 4 datasetsfull datasetsfull NOUN bracis-28371 118 5 size size NOUN bracis-28371 118 6 table table NOUN bracis-28371 118 7 3.2 3.2 NUM bracis-28371 118 8 feature feature NOUN bracis-28371 118 9 selection selection NOUN bracis-28371 118 10 each each DET bracis-28371 118 11 raw raw ADJ bracis-28371 118 12 instance instance NOUN bracis-28371 118 13 was be AUX bracis-28371 118 14 represented represent VERB bracis-28371 118 15 by by ADP bracis-28371 118 16 roughly roughly ADV bracis-28371 118 17 60 60 NUM bracis-28371 118 18 and and CCONJ bracis-28371 118 19 70 70 NUM bracis-28371 118 20 features feature NOUN bracis-28371 118 21 for for ADP bracis-28371 118 22 hmv hmv PROPN bracis-28371 118 23 and and CCONJ bracis-28371 118 24 hcpa hcpa ADJ bracis-28371 118 25 datasets dataset NOUN bracis-28371 118 26 , , PUNCT bracis-28371 118 27 respectively respectively ADV bracis-28371 118 28 . . PUNCT bracis-28371 119 1 features feature NOUN bracis-28371 119 2 with with ADP bracis-28371 119 3 more more ADJ bracis-28371 119 4 than than ADP bracis-28371 119 5 40 40 NUM bracis-28371 119 6 % % NOUN bracis-28371 119 7 of of ADP bracis-28371 119 8 missing miss VERB bracis-28371 119 9 values value NOUN bracis-28371 119 10 were be AUX bracis-28371 119 11 discarded discard VERB bracis-28371 119 12 . . PUNCT bracis-28371 120 1 to to ADP bracis-28371 120 2 pre pre VERB bracis-28371 120 3 - - ADJ bracis-28371 120 4 select select ADJ bracis-28371 120 5 features feature NOUN bracis-28371 120 6 with with ADP bracis-28371 120 7 good good ADJ bracis-28371 120 8 predictive predictive ADJ bracis-28371 120 9 power power NOUN bracis-28371 120 10 and and CCONJ bracis-28371 120 11 avoid avoid VERB bracis-28371 120 12 multicollinearity multicollinearity NOUN bracis-28371 120 13 among among ADP bracis-28371 120 14 predictive predictive ADJ bracis-28371 120 15 features feature NOUN bracis-28371 120 16 , , PUNCT bracis-28371 120 17 we we PRON bracis-28371 120 18 applied apply VERB bracis-28371 120 19 the the DET bracis-28371 120 20 correlation correlation NOUN bracis-28371 120 21 - - PUNCT bracis-28371 120 22 based base VERB bracis-28371 120 23 feature feature NOUN bracis-28371 120 24 selection selection NOUN bracis-28371 120 25 approach approach NOUN bracis-28371 120 26 ( ( PUNCT bracis-28371 120 27 cfs cfs PROPN bracis-28371 120 28 ) ) PUNCT bracis-28371 120 29 . . PUNCT bracis-28371 121 1 this this DET bracis-28371 121 2 approach approach NOUN bracis-28371 121 3 evaluates evaluate VERB bracis-28371 121 4 subsets subset NOUN bracis-28371 121 5 of of ADP bracis-28371 121 6 features feature NOUN bracis-28371 121 7 prioritizing prioritize VERB bracis-28371 121 8 features feature NOUN bracis-28371 121 9 highly highly ADV bracis-28371 121 10 correlated correlate VERB bracis-28371 121 11 with with ADP bracis-28371 121 12 the the DET bracis-28371 121 13 classes class NOUN bracis-28371 121 14 and and CCONJ bracis-28371 121 15 uncorrelated uncorrelate VERB bracis-28371 121 16 with with ADP bracis-28371 121 17 each each DET bracis-28371 121 18 other other ADJ bracis-28371 121 19     SPACE bracis-28371 122 1 [ [ X bracis-28371 122 2 7 7 NUM bracis-28371 122 3 ] ] PUNCT bracis-28371 122 4 . . PUNCT bracis-28371 123 1 feature feature NOUN bracis-28371 123 2 selection selection NOUN bracis-28371 123 3 was be AUX bracis-28371 123 4 performed perform VERB bracis-28371 123 5 only only ADV bracis-28371 123 6 over over ADP bracis-28371 123 7 the the DET bracis-28371 123 8 training training NOUN bracis-28371 123 9 instances instance NOUN bracis-28371 123 10 to to PART bracis-28371 123 11 avoid avoid VERB bracis-28371 123 12 data datum NOUN bracis-28371 123 13 leakage leakage NOUN bracis-28371 123 14 . . PUNCT bracis-28371 124 1 3.3 3.3 NUM bracis-28371 124 2 classification classification NOUN bracis-28371 124 3 we we PRON bracis-28371 124 4 applied apply VERB bracis-28371 124 5 six six NUM bracis-28371 124 6 classification classification NOUN bracis-28371 124 7 algorithms algorithm NOUN bracis-28371 124 8 : : PUNCT bracis-28371 124 9 ( ( PUNCT bracis-28371 124 10 i i NOUN bracis-28371 124 11 ) ) PUNCT bracis-28371 124 12 logistic logistic ADJ bracis-28371 124 13 regression regression NOUN bracis-28371 124 14 ( ( PUNCT bracis-28371 124 15 lr lr NOUN bracis-28371 124 16 ) ) PUNCT bracis-28371 124 17 , , PUNCT bracis-28371 124 18 three three NUM bracis-28371 124 19 tree tree NOUN bracis-28371 124 20 - - PUNCT bracis-28371 124 21 based base VERB bracis-28371 124 22 algorithms algorithm NOUN bracis-28371 124 23 , , PUNCT bracis-28371 124 24 namely namely ADV bracis-28371 124 25 ( ( PUNCT bracis-28371 124 26 ii ii NOUN bracis-28371 124 27 ) ) PUNCT bracis-28371 124 28 j48 j48 PROPN bracis-28371 124 29 , , PUNCT bracis-28371 124 30 ( ( PUNCT bracis-28371 124 31 iii iii NOUN bracis-28371 124 32 ) ) PUNCT bracis-28371 124 33 randomforest randomforest NOUN bracis-28371 124 34 ( ( PUNCT bracis-28371 124 35 rf rf NOUN bracis-28371 124 36 ) ) PUNCT bracis-28371 124 37 , , PUNCT bracis-28371 124 38 and and CCONJ bracis-28371 124 39 ( ( PUNCT bracis-28371 124 40 iv iv X bracis-28371 124 41 ) ) PUNCT bracis-28371 124 42 gradient gradient NOUN bracis-28371 124 43 boosting boost VERB bracis-28371 124 44 tree tree NOUN bracis-28371 124 45 xgboost xgboost ADV bracis-28371 124 46 ( ( PUNCT bracis-28371 124 47 xgb xgb PROPN bracis-28371 124 48 ) ) PUNCT bracis-28371 124 49 , , PUNCT bracis-28371 124 50 ( ( PUNCT bracis-28371 124 51 v v NOUN bracis-28371 124 52 ) ) PUNCT bracis-28371 124 53 a a DET bracis-28371 124 54 boosting boost VERB bracis-28371 124 55 - - PUNCT bracis-28371 124 56 based base VERB bracis-28371 124 57 ensemble ensemble ADJ bracis-28371 124 58 ( ( PUNCT bracis-28371 124 59 adaboostm1 adaboostm1 NOUN bracis-28371 124 60 ) ) PUNCT bracis-28371 124 61 , , PUNCT bracis-28371 124 62 and and CCONJ bracis-28371 124 63 ( ( PUNCT bracis-28371 124 64 vi vi NOUN bracis-28371 124 65 ) ) PUNCT bracis-28371 124 66 multi multi ADJ bracis-28371 124 67 layer layer NOUN bracis-28371 124 68 perceptron perceptron PROPN bracis-28371 124 69 ( ( PUNCT bracis-28371 124 70 mlp mlp PROPN bracis-28371 124 71 ) ) PUNCT bracis-28371 124 72 . . PUNCT bracis-28371 125 1 in in ADP bracis-28371 125 2 supervised supervised ADJ bracis-28371 125 3 classification classification NOUN bracis-28371 125 4 approaches approach NOUN bracis-28371 125 5 , , PUNCT bracis-28371 125 6 class class NOUN bracis-28371 125 7 imbalance imbalance NOUN bracis-28371 125 8 refers refer VERB bracis-28371 125 9 to to ADP bracis-28371 125 10 the the DET bracis-28371 125 11 inequality inequality NOUN bracis-28371 125 12 of of ADP bracis-28371 125 13 the the DET bracis-28371 125 14 proportions proportion NOUN bracis-28371 125 15 of of ADP bracis-28371 125 16 instances instance NOUN bracis-28371 125 17 for for ADP bracis-28371 125 18 each each DET bracis-28371 125 19 class class NOUN bracis-28371 125 20 . . PUNCT bracis-28371 126 1 class class NOUN bracis-28371 126 2 imbalance imbalance NOUN bracis-28371 126 3 is be AUX bracis-28371 126 4 frequently frequently ADV bracis-28371 126 5 observed observe VERB bracis-28371 126 6 in in ADP bracis-28371 126 7 healthcare healthcare PROPN bracis-28371 126 8 data datum NOUN bracis-28371 126 9 , , PUNCT bracis-28371 126 10 where where SCONJ bracis-28371 126 11 the the DET bracis-28371 126 12 class class NOUN bracis-28371 126 13 of of ADP bracis-28371 126 14 interest interest NOUN bracis-28371 126 15 is be AUX bracis-28371 126 16 often often ADV bracis-28371 126 17 the the DET bracis-28371 126 18 minority minority NOUN bracis-28371 126 19 ( ( PUNCT bracis-28371 126 20 e.g. e.g. ADV bracis-28371 126 21 , , PUNCT bracis-28371 126 22 disease disease NOUN bracis-28371 126 23 occurrence occurrence NOUN bracis-28371 126 24 or or CCONJ bracis-28371 126 25 poor poor ADJ bracis-28371 126 26 diagnosis diagnosis NOUN bracis-28371 126 27 ) ) PUNCT bracis-28371 126 28 . . PUNCT bracis-28371 127 1 when when SCONJ bracis-28371 127 2 training train VERB bracis-28371 127 3 classification classification NOUN bracis-28371 127 4 algorithms algorithm NOUN bracis-28371 127 5 on on ADP bracis-28371 127 6 imbalanced imbalanced ADJ bracis-28371 127 7 data datum NOUN bracis-28371 127 8 without without ADP bracis-28371 127 9 the the DET bracis-28371 127 10 necessary necessary ADJ bracis-28371 127 11 care care NOUN bracis-28371 127 12 , , PUNCT bracis-28371 127 13 these these DET bracis-28371 127 14 algorithms algorithm NOUN bracis-28371 127 15 tend tend VERB bracis-28371 127 16 to to PART bracis-28371 127 17 present present VERB bracis-28371 127 18 biased biased ADJ bracis-28371 127 19 results result NOUN bracis-28371 127 20 , , PUNCT bracis-28371 127 21 prioritizing prioritize VERB bracis-28371 127 22 the the DET bracis-28371 127 23 performance performance NOUN bracis-28371 127 24 of of ADP bracis-28371 127 25 the the DET bracis-28371 127 26 majority majority NOUN bracis-28371 127 27 class class NOUN bracis-28371 127 28 . . PUNCT bracis-28371 128 1 in in ADP bracis-28371 128 2 our our PRON bracis-28371 128 3 datasets dataset NOUN bracis-28371 128 4 , , PUNCT bracis-28371 128 5 the the DET bracis-28371 128 6 outcome outcome NOUN bracis-28371 128 7 variable variable ADJ bracis-28371 128 8 presents present NOUN bracis-28371 128 9 , , PUNCT bracis-28371 128 10 to to ADP bracis-28371 128 11 some some DET bracis-28371 128 12 degree degree NOUN bracis-28371 128 13 , , PUNCT bracis-28371 128 14 an an DET bracis-28371 128 15 imbalance imbalance NOUN bracis-28371 128 16 between between ADP bracis-28371 128 17 the the DET bracis-28371 128 18 classes class NOUN bracis-28371 128 19 ( ( PUNCT bracis-28371 128 20 alive alive ADJ bracis-28371 128 21 and and CCONJ bracis-28371 128 22 deceased deceased ADJ bracis-28371 128 23 ) ) PUNCT bracis-28371 128 24 . . PUNCT bracis-28371 129 1 the the DET bracis-28371 129 2 greatest great ADJ bracis-28371 129 3 imbalance imbalance NOUN bracis-28371 129 4 is be AUX bracis-28371 129 5 observed observe VERB bracis-28371 129 6 in in ADP bracis-28371 129 7 the the DET bracis-28371 129 8 \(\text \(\text NOUN bracis-28371 129 9 { { PUNCT bracis-28371 129 10 hmv}_{all}\ hmv}_{all}\ PROPN bracis-28371 129 11 ) ) PUNCT bracis-28371 129 12 dataset dataset NOUN bracis-28371 129 13 , , PUNCT bracis-28371 129 14 with with ADP bracis-28371 129 15 approximately approximately ADV bracis-28371 129 16 7.38 7.38 NUM bracis-28371 129 17 patients patient NOUN bracis-28371 129 18 with with ADP bracis-28371 129 19 an an DET bracis-28371 129 20 alive alive ADJ bracis-28371 129 21 outcome outcome NOUN bracis-28371 129 22 for for ADP bracis-28371 129 23 each each DET bracis-28371 129 24 patient patient NOUN bracis-28371 129 25 with with ADP bracis-28371 129 26 a a DET bracis-28371 129 27 deceased deceased ADJ bracis-28371 129 28 outcome outcome NOUN bracis-28371 129 29 . . PUNCT bracis-28371 130 1 considering consider VERB bracis-28371 130 2 only only ADV bracis-28371 130 3 patients patient NOUN bracis-28371 130 4 admitted admit VERB bracis-28371 130 5 to to ADP bracis-28371 130 6 the the DET bracis-28371 130 7 icu icu PROPN bracis-28371 130 8 , , PUNCT bracis-28371 130 9 the the DET bracis-28371 130 10 imbalance imbalance NOUN bracis-28371 130 11 is be AUX bracis-28371 130 12 2.2 2.2 NUM bracis-28371 130 13 and and CCONJ bracis-28371 130 14 1.67 1.67 NUM bracis-28371 130 15 alive alive ADJ bracis-28371 130 16 for for ADP bracis-28371 130 17 each each DET bracis-28371 130 18 deceased deceased ADJ bracis-28371 130 19 , , PUNCT bracis-28371 130 20 for for ADP bracis-28371 130 21 the the DET bracis-28371 130 22 \(\text \(\text NOUN bracis-28371 130 23 { { PUNCT bracis-28371 130 24 hmv}_{cti}\ hmv}_{cti}\ PROPN bracis-28371 130 25 ) ) PUNCT bracis-28371 130 26 and and CCONJ bracis-28371 130 27 \(\text \(\text NOUN bracis-28371 130 28 { { PUNCT bracis-28371 130 29 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 130 30 ) ) PUNCT bracis-28371 130 31 datasets dataset NOUN bracis-28371 130 32 , , PUNCT bracis-28371 130 33 respectively respectively ADV bracis-28371 130 34 . . PUNCT bracis-28371 131 1 to to PART bracis-28371 131 2 deal deal VERB bracis-28371 131 3 with with ADP bracis-28371 131 4 class class NOUN bracis-28371 131 5 imbalance imbalance NOUN bracis-28371 131 6 , , PUNCT bracis-28371 131 7 we we PRON bracis-28371 131 8 applied apply VERB bracis-28371 131 9 the the DET bracis-28371 131 10 cost cost NOUN bracis-28371 131 11 - - PUNCT bracis-28371 131 12 sensitive sensitive ADJ bracis-28371 131 13 matrix matrix NOUN bracis-28371 131 14 approach approach NOUN bracis-28371 131 15 , , PUNCT bracis-28371 131 16 in in ADP bracis-28371 131 17 which which PRON bracis-28371 131 18 the the DET bracis-28371 131 19 classification classification NOUN bracis-28371 131 20 errors error NOUN bracis-28371 131 21 ( ( PUNCT bracis-28371 131 22 false false ADJ bracis-28371 131 23 - - PUNCT bracis-28371 131 24 negatives negative NOUN bracis-28371 131 25 ( ( PUNCT bracis-28371 131 26 fn fn NOUN bracis-28371 131 27 ) ) PUNCT bracis-28371 131 28 or or CCONJ bracis-28371 131 29 false false ADJ bracis-28371 131 30 - - PUNCT bracis-28371 131 31 positives positive NOUN bracis-28371 131 32 ( ( PUNCT bracis-28371 131 33 fp fp NOUN bracis-28371 131 34 ) ) PUNCT bracis-28371 131 35 ) ) PUNCT bracis-28371 131 36 are be AUX bracis-28371 131 37 weighted weight VERB bracis-28371 131 38 depending depend VERB bracis-28371 131 39 on on ADP bracis-28371 131 40 the the DET bracis-28371 131 41 task task NOUN bracis-28371 131 42 to to PART bracis-28371 131 43 be be AUX bracis-28371 131 44 addressed address VERB bracis-28371 131 45 [ [ X bracis-28371 131 46 14 14 NUM bracis-28371 131 47 ] ] PUNCT bracis-28371 131 48 . . PUNCT bracis-28371 132 1 since since SCONJ bracis-28371 132 2 the the DET bracis-28371 132 3 \(outcome \(outcome NOUN bracis-28371 132 4 = = NOUN bracis-28371 132 5 deceased\ deceased\ X bracis-28371 132 6 ) ) PUNCT bracis-28371 132 7 is be AUX bracis-28371 132 8 our our PRON bracis-28371 132 9 positive positive ADJ bracis-28371 132 10 class class NOUN bracis-28371 132 11 , , PUNCT bracis-28371 132 12 we we PRON bracis-28371 132 13 applied apply VERB bracis-28371 132 14 a a DET bracis-28371 132 15 cost cost NOUN bracis-28371 132 16 matrix matrix NOUN bracis-28371 132 17 that that PRON bracis-28371 132 18 increased increase VERB bracis-28371 132 19 the the DET bracis-28371 132 20 penalty penalty NOUN bracis-28371 132 21 for for ADP bracis-28371 132 22 fn fn PROPN bracis-28371 132 23 . . PUNCT bracis-28371 133 1 the the DET bracis-28371 133 2 cost cost NOUN bracis-28371 133 3 matrices matrix NOUN bracis-28371 133 4 applied apply VERB bracis-28371 133 5 considered consider VERB bracis-28371 133 6 the the DET bracis-28371 133 7 imbalance imbalance NOUN bracis-28371 133 8 between between ADP bracis-28371 133 9 classes class NOUN bracis-28371 133 10 so so SCONJ bracis-28371 133 11 that that SCONJ bracis-28371 133 12 the the DET bracis-28371 133 13 penalty penalty NOUN bracis-28371 133 14 for for ADP bracis-28371 133 15 false false ADJ bracis-28371 133 16 negatives negative NOUN bracis-28371 133 17 was be AUX bracis-28371 133 18 defined define VERB bracis-28371 133 19 as as ADP bracis-28371 133 20 \(pen \(pen PROPN bracis-28371 133 21 = = PUNCT bracis-28371 133 22 \frac{nc\_maj}{nc\_min}\ \frac{nc\_maj}{nc\_min}\ PROPN bracis-28371 133 23 ) ) PUNCT bracis-28371 133 24 where where SCONJ bracis-28371 133 25 \(nc\_maj\ \(nc\_maj\ NOUN bracis-28371 133 26 ) ) PUNCT bracis-28371 133 27 is be AUX bracis-28371 133 28 the the DET bracis-28371 133 29 number number NOUN bracis-28371 133 30 of of ADP bracis-28371 133 31 instances instance NOUN bracis-28371 133 32 of of ADP bracis-28371 133 33 the the DET bracis-28371 133 34 majority majority NOUN bracis-28371 133 35 class class NOUN bracis-28371 133 36 and and CCONJ bracis-28371 133 37 \(nc\_min\ \(nc\_min\ NOUN bracis-28371 133 38 ) ) PUNCT bracis-28371 133 39 is be AUX bracis-28371 133 40 the the DET bracis-28371 133 41 number number NOUN bracis-28371 133 42 of of ADP bracis-28371 133 43 instances instance NOUN bracis-28371 133 44 of of ADP bracis-28371 133 45 the the DET bracis-28371 133 46 minority minority NOUN bracis-28371 133 47 class class NOUN bracis-28371 133 48 . . PUNCT bracis-28371 134 1 3.4 3.4 NUM bracis-28371 134 2 evaluation evaluation NOUN bracis-28371 134 3 metrics metric NOUN bracis-28371 134 4 . . PUNCT bracis-28371 135 1 with with ADP bracis-28371 135 2 the the DET bracis-28371 135 3 aim aim NOUN bracis-28371 135 4 of of ADP bracis-28371 135 5 evaluating evaluate VERB bracis-28371 135 6 the the DET bracis-28371 135 7 classification classification NOUN bracis-28371 135 8 models model NOUN bracis-28371 135 9 , , PUNCT bracis-28371 135 10 we we PRON bracis-28371 135 11 calculated calculate VERB bracis-28371 135 12 several several ADJ bracis-28371 135 13 metrics metric NOUN bracis-28371 135 14 , , PUNCT bracis-28371 135 15 namely namely ADV bracis-28371 135 16 : : PUNCT bracis-28371 135 17 sensitivity sensitivity NOUN bracis-28371 135 18 ( ( PUNCT bracis-28371 135 19 also also ADV bracis-28371 135 20 known know VERB bracis-28371 135 21 as as ADP bracis-28371 135 22 recall recall NOUN bracis-28371 135 23 or or CCONJ bracis-28371 135 24 true true ADJ bracis-28371 135 25 positive positive ADJ bracis-28371 135 26 rate rate NOUN bracis-28371 135 27 ) ) PUNCT bracis-28371 135 28 , , PUNCT bracis-28371 135 29 specificity specificity NOUN bracis-28371 135 30 ( ( PUNCT bracis-28371 135 31 or or CCONJ bracis-28371 135 32 true true ADJ bracis-28371 135 33 negative negative ADJ bracis-28371 135 34 rate rate NOUN bracis-28371 135 35 ) ) PUNCT bracis-28371 135 36 , , PUNCT bracis-28371 135 37 positive positive ADJ bracis-28371 135 38 predictive predictive ADJ bracis-28371 135 39 value value NOUN bracis-28371 135 40 ( ( PUNCT bracis-28371 135 41 ppv ppv NOUN bracis-28371 135 42 or or CCONJ bracis-28371 135 43 precision precision NOUN bracis-28371 135 44 ) ) PUNCT bracis-28371 135 45 , , PUNCT bracis-28371 135 46 negative negative ADJ bracis-28371 135 47 predictive predictive ADJ bracis-28371 135 48 value value NOUN bracis-28371 135 49 ( ( PUNCT bracis-28371 135 50 npv npv NOUN bracis-28371 135 51 ) ) PUNCT bracis-28371 135 52 , , PUNCT bracis-28371 135 53 f1 f1 NOUN bracis-28371 135 54 ( ( PUNCT bracis-28371 135 55 both both PRON bracis-28371 135 56 for for ADP bracis-28371 135 57 the the DET bracis-28371 135 58 positive positive ADJ bracis-28371 135 59 class class NOUN bracis-28371 135 60 ( ( PUNCT bracis-28371 135 61 f1\(_+\ f1\(_+\ ADJ bracis-28371 135 62 ) ) PUNCT bracis-28371 135 63 ) ) PUNCT bracis-28371 135 64 and and CCONJ bracis-28371 135 65 macro macro VERB bracis-28371 135 66 - - ADJ bracis-28371 135 67 averaged averaged ADJ bracis-28371 135 68 ( ( PUNCT bracis-28371 135 69 ma ma NOUN bracis-28371 135 70 - - PUNCT bracis-28371 135 71 f1 f1 NOUN bracis-28371 135 72 ) ) PUNCT bracis-28371 135 73 for for ADP bracis-28371 135 74 both both DET bracis-28371 135 75 classes class NOUN bracis-28371 135 76 ) ) PUNCT bracis-28371 135 77 , , PUNCT bracis-28371 135 78 cohen cohen PROPN bracis-28371 135 79 ’s ’s PART bracis-28371 135 80 kappa kappa PROPN bracis-28371 135 81 , , PUNCT bracis-28371 135 82 area area NOUN bracis-28371 135 83 under under ADP bracis-28371 135 84 receiver receiver NOUN bracis-28371 135 85 operating operate VERB bracis-28371 135 86 characteristic characteristic ADJ bracis-28371 135 87 curve curve NOUN bracis-28371 135 88 ( ( PUNCT bracis-28371 135 89 auroc auroc NOUN bracis-28371 135 90 ) ) PUNCT bracis-28371 135 91 , , PUNCT bracis-28371 135 92 and and CCONJ bracis-28371 135 93 area area NOUN bracis-28371 135 94 under under ADP bracis-28371 135 95 precision precision NOUN bracis-28371 135 96 - - PUNCT bracis-28371 135 97 recall recall NOUN bracis-28371 135 98 curve curve NOUN bracis-28371 135 99 ( ( PUNCT bracis-28371 135 100 auprc auprc ADJ bracis-28371 135 101 ) ) PUNCT bracis-28371 135 102 . . PUNCT bracis-28371 136 1 using use VERB bracis-28371 136 2 different different ADJ bracis-28371 136 3 metrics metric NOUN bracis-28371 136 4 is be AUX bracis-28371 136 5 important important ADJ bracis-28371 136 6 to to PART bracis-28371 136 7 interpret interpret VERB bracis-28371 136 8 the the DET bracis-28371 136 9 results result NOUN bracis-28371 136 10 from from ADP bracis-28371 136 11 different different ADJ bracis-28371 136 12 perspectives perspective NOUN bracis-28371 136 13 . . PUNCT bracis-28371 137 1 resampling resample VERB bracis-28371 137 2 . . PUNCT bracis-28371 138 1 we we PRON bracis-28371 138 2 used use VERB bracis-28371 138 3 the the DET bracis-28371 138 4 resampling resample VERB bracis-28371 138 5 technique technique NOUN bracis-28371 138 6 bootstrap bootstrap NOUN bracis-28371 138 7 to to PART bracis-28371 138 8 evaluate evaluate VERB bracis-28371 138 9 the the DET bracis-28371 138 10 predictive predictive ADJ bracis-28371 138 11 models model NOUN bracis-28371 138 12 . . PUNCT bracis-28371 139 1 for for ADP bracis-28371 139 2 each each DET bracis-28371 139 3 dataset dataset NOUN bracis-28371 139 4 , , PUNCT bracis-28371 139 5 we we PRON bracis-28371 139 6 performed perform VERB bracis-28371 139 7 100 100 NUM bracis-28371 139 8 iterations iteration NOUN bracis-28371 139 9 of of ADP bracis-28371 139 10 resampling resample VERB bracis-28371 139 11 with with ADP bracis-28371 139 12 replacement replacement NOUN bracis-28371 139 13 . . PUNCT bracis-28371 140 1 for for ADP bracis-28371 140 2 each each DET bracis-28371 140 3 iteration iteration NOUN bracis-28371 140 4 , , PUNCT bracis-28371 140 5 several several ADJ bracis-28371 140 6 instances instance NOUN bracis-28371 140 7 — — PUNCT bracis-28371 140 8 equal equal ADJ bracis-28371 140 9 to to ADP bracis-28371 140 10 the the DET bracis-28371 140 11 number number NOUN bracis-28371 140 12 of of ADP bracis-28371 140 13 instances instance NOUN bracis-28371 140 14 in in ADP bracis-28371 140 15 the the DET bracis-28371 140 16 dataset dataset NOUN bracis-28371 140 17 — — PUNCT bracis-28371 140 18 were be AUX bracis-28371 140 19 randomly randomly ADV bracis-28371 140 20 selected select VERB bracis-28371 140 21 to to PART bracis-28371 140 22 compose compose VERB bracis-28371 140 23 the the DET bracis-28371 140 24 training training NOUN bracis-28371 140 25 set set NOUN bracis-28371 140 26 , , PUNCT bracis-28371 140 27 and and CCONJ bracis-28371 140 28 the the DET bracis-28371 140 29 remaining remain VERB bracis-28371 140 30 instances instance NOUN bracis-28371 140 31 to to PART bracis-28371 140 32 constitute constitute VERB bracis-28371 140 33 the the DET bracis-28371 140 34 test test NOUN bracis-28371 140 35 set set VERB bracis-28371 140 36 . . PUNCT bracis-28371 141 1 thus thus ADV bracis-28371 141 2 , , PUNCT bracis-28371 141 3 for for ADP bracis-28371 141 4 each each PRON bracis-28371 141 5 of of ADP bracis-28371 141 6 the the DET bracis-28371 141 7 100 100 NUM bracis-28371 141 8 iterations iteration NOUN bracis-28371 141 9 , , PUNCT bracis-28371 141 10 the the DET bracis-28371 141 11 classifiers classifier NOUN bracis-28371 141 12 were be AUX bracis-28371 141 13 trained train VERB bracis-28371 141 14 on on ADP bracis-28371 141 15 the the DET bracis-28371 141 16 training training NOUN bracis-28371 141 17 set set NOUN bracis-28371 141 18 and and CCONJ bracis-28371 141 19 evaluated evaluate VERB bracis-28371 141 20 on on ADP bracis-28371 141 21 the the DET bracis-28371 141 22 test test NOUN bracis-28371 141 23 set set VERB bracis-28371 141 24 . . PUNCT bracis-28371 142 1 at at ADP bracis-28371 142 2 the the DET bracis-28371 142 3 end end NOUN bracis-28371 142 4 of of ADP bracis-28371 142 5 the the DET bracis-28371 142 6 process process NOUN bracis-28371 142 7 , , PUNCT bracis-28371 142 8 we we PRON bracis-28371 142 9 take take VERB bracis-28371 142 10 the the DET bracis-28371 142 11 means mean NOUN bracis-28371 142 12 of of ADP bracis-28371 142 13 the the DET bracis-28371 142 14 evaluation evaluation NOUN bracis-28371 142 15 metrics metric NOUN bracis-28371 142 16 . . PUNCT bracis-28371 143 1 3.5 3.5 NUM bracis-28371 143 2 explainability explainability NOUN bracis-28371 143 3 of of ADP bracis-28371 143 4     SPACE bracis-28371 143 5 classification classification NOUN bracis-28371 143 6 results result NOUN bracis-28371 143 7 this this DET bracis-28371 143 8 section section NOUN bracis-28371 143 9 describes describe VERB bracis-28371 143 10 the the DET bracis-28371 143 11 proposed propose VERB bracis-28371 143 12 approach approach NOUN bracis-28371 143 13 to to PART bracis-28371 143 14 help help VERB bracis-28371 143 15 explain explain VERB bracis-28371 143 16 the the DET bracis-28371 143 17 features feature NOUN bracis-28371 143 18 taken take VERB bracis-28371 143 19 into into ADP bracis-28371 143 20 account account NOUN bracis-28371 143 21 by by ADP bracis-28371 143 22 the the DET bracis-28371 143 23 underlying underlie VERB bracis-28371 143 24 model model NOUN bracis-28371 143 25 during during ADP bracis-28371 143 26 inference inference NOUN bracis-28371 143 27 . . PUNCT bracis-28371 144 1 the the DET bracis-28371 144 2 goal goal NOUN bracis-28371 144 3 is be AUX bracis-28371 144 4 to to PART bracis-28371 144 5 extract extract VERB bracis-28371 144 6 insights insight NOUN bracis-28371 144 7 into into ADP bracis-28371 144 8 the the DET bracis-28371 144 9 relationship relationship NOUN bracis-28371 144 10 between between ADP bracis-28371 144 11 the the DET bracis-28371 144 12 features feature NOUN bracis-28371 144 13 utilized utilize VERB bracis-28371 144 14 by by ADP bracis-28371 144 15 the the DET bracis-28371 144 16 predictive predictive ADJ bracis-28371 144 17 models model NOUN bracis-28371 144 18 and and CCONJ bracis-28371 144 19 their their PRON bracis-28371 144 20 correct correct ADJ bracis-28371 144 21 and and CCONJ bracis-28371 144 22 incorrect incorrect ADJ bracis-28371 144 23 predictions prediction NOUN bracis-28371 144 24 . . PUNCT bracis-28371 145 1 the the DET bracis-28371 145 2 proposed propose VERB bracis-28371 145 3 approach approach NOUN bracis-28371 145 4 corresponds correspond VERB bracis-28371 145 5 to to PART bracis-28371 145 6 step step VERB bracis-28371 145 7 four four NUM bracis-28371 145 8 , , PUNCT bracis-28371 145 9 shown show VERB bracis-28371 145 10 in in ADP bracis-28371 145 11 fig fig NOUN bracis-28371 145 12 . . PUNCT bracis-28371 145 13     SPACE bracis-28371 146 1 1 1 NUM bracis-28371 146 2 and and CCONJ bracis-28371 146 3 its its PRON bracis-28371 146 4 components component NOUN bracis-28371 146 5 are be AUX bracis-28371 146 6 shown show VERB bracis-28371 146 7 in in ADP bracis-28371 146 8 fig fig NOUN bracis-28371 146 9 . . PUNCT bracis-28371 146 10     SPACE bracis-28371 147 1 2 2 X bracis-28371 147 2 . . PUNCT bracis-28371 148 1 the the DET bracis-28371 148 2 idea idea NOUN bracis-28371 148 3 is be AUX bracis-28371 148 4 similar similar ADJ bracis-28371 148 5 to to ADP bracis-28371 148 6 supervised supervised ADJ bracis-28371 148 7 clustering clustering NOUN bracis-28371 148 8     SPACE bracis-28371 149 1 [ [ X bracis-28371 149 2 10 10 NUM bracis-28371 149 3 ] ] PUNCT bracis-28371 149 4 , , PUNCT bracis-28371 149 5 in in ADP bracis-28371 149 6 which which PRON bracis-28371 149 7 shap shap PROPN bracis-28371 149 8 values value NOUN bracis-28371 149 9 ( ( PUNCT bracis-28371 149 10 not not PART bracis-28371 149 11 raw raw ADJ bracis-28371 149 12 feature feature NOUN bracis-28371 149 13 values value NOUN bracis-28371 149 14 ) ) PUNCT bracis-28371 149 15 are be AUX bracis-28371 149 16 used use VERB bracis-28371 149 17 to to PART bracis-28371 149 18 cluster cluster NOUN bracis-28371 149 19 instances instance NOUN bracis-28371 149 20 . . PUNCT bracis-28371 150 1 however however ADV bracis-28371 150 2 , , PUNCT bracis-28371 150 3 instead instead ADV bracis-28371 150 4 of of ADP bracis-28371 150 5 clustering clustering NOUN bracis-28371 150 6 , , PUNCT bracis-28371 150 7 we we PRON bracis-28371 150 8 use use VERB bracis-28371 150 9 them they PRON bracis-28371 150 10 to to PART bracis-28371 150 11 generate generate VERB bracis-28371 150 12 visual visual ADJ bracis-28371 150 13 representations representation NOUN bracis-28371 150 14 to to PART bracis-28371 150 15 inspect inspect VERB bracis-28371 150 16 predictions prediction NOUN bracis-28371 150 17 . . PUNCT bracis-28371 151 1 fig fig NOUN bracis-28371 151 2 . . PUNCT bracis-28371 152 1 2 2 X bracis-28371 152 2 . . X bracis-28371 152 3 steps step NOUN bracis-28371 152 4 applied apply VERB bracis-28371 152 5 in in ADP bracis-28371 152 6 explaining explain VERB bracis-28371 152 7 and and CCONJ bracis-28371 152 8 interpreting interpret VERB bracis-28371 152 9 the the DET bracis-28371 152 10 results result NOUN bracis-28371 152 11 returned return VERB bracis-28371 152 12 by by ADP bracis-28371 152 13 the the DET bracis-28371 152 14 predictive predictive ADJ bracis-28371 152 15 models model NOUN bracis-28371 152 16 . . PUNCT bracis-28371 153 1 full full ADJ bracis-28371 153 2 size size NOUN bracis-28371 153 3 image image NOUN bracis-28371 153 4 first first ADV bracis-28371 153 5 , , PUNCT bracis-28371 153 6 starting start VERB bracis-28371 153 7 from from ADP bracis-28371 153 8 a a DET bracis-28371 153 9 generic generic ADJ bracis-28371 153 10 dataset dataset NOUN bracis-28371 153 11 d d NOUN bracis-28371 153 12 , , PUNCT bracis-28371 153 13 which which PRON bracis-28371 153 14 in in ADP bracis-28371 153 15 this this DET bracis-28371 153 16 case case NOUN bracis-28371 153 17 , , PUNCT bracis-28371 153 18 could could AUX bracis-28371 153 19 be be AUX bracis-28371 153 20 any any PRON bracis-28371 153 21 of of ADP bracis-28371 153 22 our our PRON bracis-28371 153 23 datasets dataset NOUN bracis-28371 153 24 , , PUNCT bracis-28371 153 25 we we PRON bracis-28371 153 26 use use VERB bracis-28371 153 27 dimensionality dimensionality NOUN bracis-28371 153 28 reduction reduction NOUN bracis-28371 153 29 to to PART bracis-28371 153 30 understand understand VERB bracis-28371 153 31 aspects aspect NOUN bracis-28371 153 32 of of ADP bracis-28371 153 33 linear linear ADJ bracis-28371 153 34 separability separability NOUN bracis-28371 153 35 of of ADP bracis-28371 153 36 the the DET bracis-28371 153 37 data datum NOUN bracis-28371 153 38 . . PUNCT bracis-28371 154 1 it it PRON bracis-28371 154 2 is be AUX bracis-28371 154 3 expected expect VERB bracis-28371 154 4 that that SCONJ bracis-28371 154 5 in in ADP bracis-28371 154 6 the the DET bracis-28371 154 7 raw raw ADJ bracis-28371 154 8 data datum NOUN bracis-28371 154 9 , , PUNCT bracis-28371 154 10 there there PRON bracis-28371 154 11 are be VERB bracis-28371 154 12 no no DET bracis-28371 154 13 clear clear ADJ bracis-28371 154 14 separability separability NOUN bracis-28371 154 15 patterns pattern NOUN bracis-28371 154 16 . . PUNCT bracis-28371 155 1 rectangles rectangle VERB bracis-28371 155 2     SPACE bracis-28371 155 3 2.a 2.a NUM bracis-28371 155 4 and and CCONJ bracis-28371 155 5     SPACE bracis-28371 155 6 3.a 3.a NUM bracis-28371 155 7 represent represent VERB bracis-28371 155 8 the the DET bracis-28371 155 9 visualization visualization NOUN bracis-28371 155 10 of of ADP bracis-28371 155 11 the the DET bracis-28371 155 12 raw raw ADJ bracis-28371 155 13 data datum NOUN bracis-28371 155 14 using use VERB bracis-28371 155 15 a a DET bracis-28371 155 16 t t PROPN bracis-28371 155 17 - - PUNCT bracis-28371 155 18 sne sne NOUN bracis-28371 155 19 projection projection NOUN bracis-28371 155 20     SPACE bracis-28371 156 1 [ [ X bracis-28371 156 2 12 12 NUM bracis-28371 156 3 ] ] PUNCT bracis-28371 156 4 . . PUNCT bracis-28371 157 1 we we PRON bracis-28371 157 2 can can AUX bracis-28371 157 3 see see VERB bracis-28371 157 4 that that SCONJ bracis-28371 157 5 the the DET bracis-28371 157 6 arrangement arrangement NOUN bracis-28371 157 7 of of ADP bracis-28371 157 8 instances instance NOUN bracis-28371 157 9 does do AUX bracis-28371 157 10 not not PART bracis-28371 157 11 constitute constitute VERB bracis-28371 157 12 clusters cluster NOUN bracis-28371 157 13 , , PUNCT bracis-28371 157 14 suggesting suggest VERB bracis-28371 157 15 little little ADJ bracis-28371 157 16 linearity linearity NOUN bracis-28371 157 17 in in ADP bracis-28371 157 18 the the DET bracis-28371 157 19 data datum NOUN bracis-28371 157 20 . . PUNCT bracis-28371 158 1 starting start VERB bracis-28371 158 2 from from ADP bracis-28371 158 3 the the DET bracis-28371 158 4 same same ADJ bracis-28371 158 5 dataset dataset NOUN bracis-28371 158 6 d d NOUN bracis-28371 158 7 , , PUNCT bracis-28371 158 8 the the DET bracis-28371 158 9 proposed propose VERB bracis-28371 158 10 approach approach NOUN bracis-28371 158 11 consists consist VERB bracis-28371 158 12 of of ADP bracis-28371 158 13 the the DET bracis-28371 158 14 following follow VERB bracis-28371 158 15 steps step NOUN bracis-28371 158 16 : : PUNCT bracis-28371 158 17 ( ( PUNCT bracis-28371 158 18 2.b 2.b NUM bracis-28371 158 19 ) ) PUNCT bracis-28371 158 20 training train VERB bracis-28371 158 21 a a DET bracis-28371 158 22 classification classification NOUN bracis-28371 158 23 algorithm algorithm NOUN bracis-28371 158 24 and and CCONJ bracis-28371 158 25 ( ( PUNCT bracis-28371 158 26 3.b 3.b NUM bracis-28371 158 27 ) ) PUNCT bracis-28371 158 28 applying apply VERB bracis-28371 158 29 the the DET bracis-28371 158 30 shap shap NOUN bracis-28371 158 31 explanation explanation NOUN bracis-28371 158 32 over over ADP bracis-28371 158 33 predictions prediction NOUN bracis-28371 158 34 . . PUNCT bracis-28371 159 1 the the DET bracis-28371 159 2 training training NOUN bracis-28371 159 3 process process NOUN bracis-28371 159 4 in in ADP bracis-28371 159 5 2.b 2.b PROPN bracis-28371 159 6 follows follow VERB bracis-28371 159 7 standard standard ADJ bracis-28371 159 8 practices practice NOUN bracis-28371 159 9 , , PUNCT bracis-28371 159 10 such such ADJ bracis-28371 159 11 as as ADP bracis-28371 159 12 handling handle VERB bracis-28371 159 13 missing miss VERB bracis-28371 159 14 values value NOUN bracis-28371 159 15 , , PUNCT bracis-28371 159 16 applying apply VERB bracis-28371 159 17 techniques technique NOUN bracis-28371 159 18 to to PART bracis-28371 159 19 deal deal VERB bracis-28371 159 20 with with ADP bracis-28371 159 21 class class NOUN bracis-28371 159 22 imbalance imbalance NOUN bracis-28371 159 23 , , PUNCT bracis-28371 159 24 model model NOUN bracis-28371 159 25 evaluation evaluation NOUN bracis-28371 159 26 , , PUNCT bracis-28371 159 27 and and CCONJ bracis-28371 159 28 calibration calibration NOUN bracis-28371 159 29 . . PUNCT bracis-28371 160 1 in in ADP bracis-28371 160 2 3.b 3.b NUM bracis-28371 160 3 , , PUNCT bracis-28371 160 4 we we PRON bracis-28371 160 5 select select VERB bracis-28371 160 6 instances instance NOUN bracis-28371 160 7 from from ADP bracis-28371 160 8 the the DET bracis-28371 160 9 test test NOUN bracis-28371 160 10 set set VERB bracis-28371 160 11 and and CCONJ bracis-28371 160 12 submit submit VERB bracis-28371 160 13 them they PRON bracis-28371 160 14 to to ADP bracis-28371 160 15 the the DET bracis-28371 160 16 shap shap PROPN bracis-28371 160 17 approach approach NOUN bracis-28371 160 18 . . PUNCT bracis-28371 161 1 we we PRON bracis-28371 161 2 obtain obtain VERB bracis-28371 161 3 , , PUNCT bracis-28371 161 4 for for ADP bracis-28371 161 5 each each DET bracis-28371 161 6 feature feature NOUN bracis-28371 161 7 value value NOUN bracis-28371 161 8 v v NOUN bracis-28371 161 9 of of ADP bracis-28371 161 10 the the DET bracis-28371 161 11 instance instance NOUN bracis-28371 162 1 i i PRON bracis-28371 162 2 , , PUNCT bracis-28371 162 3 their their PRON bracis-28371 162 4 contribution contribution NOUN bracis-28371 162 5 values value NOUN bracis-28371 162 6 ( ( PUNCT bracis-28371 162 7 shap shap NOUN bracis-28371 162 8 values value NOUN bracis-28371 162 9 ) ) PUNCT bracis-28371 162 10 . . PUNCT bracis-28371 163 1 in in ADP bracis-28371 163 2 our our PRON bracis-28371 163 3 experiments experiment NOUN bracis-28371 163 4 , , PUNCT bracis-28371 163 5 we we PRON bracis-28371 163 6 used use VERB bracis-28371 163 7 kernelshap kernelshap PROPN bracis-28371 163 8 to to PART bracis-28371 163 9 estimate estimate VERB bracis-28371 163 10 shap shap PROPN bracis-28371 163 11 values value NOUN bracis-28371 163 12 , , PUNCT bracis-28371 163 13 but but CCONJ bracis-28371 163 14 other other ADJ bracis-28371 163 15 methods method NOUN bracis-28371 163 16 could could AUX bracis-28371 163 17 also also ADV bracis-28371 163 18 have have AUX bracis-28371 163 19 been be AUX bracis-28371 163 20 used use VERB bracis-28371 163 21 ( ( PUNCT bracis-28371 163 22 i.e. i.e. X bracis-28371 163 23 , , PUNCT bracis-28371 163 24 treeshap treeshap ADJ bracis-28371 163 25 , , PUNCT bracis-28371 163 26 linear linear PROPN bracis-28371 163 27 shap shap PROPN bracis-28371 164 1 [ [ X bracis-28371 164 2 11 11 NUM bracis-28371 164 3 ] ] PUNCT bracis-28371 164 4 , , PUNCT bracis-28371 164 5 kernel kernel PROPN bracis-28371 164 6 shap shap PROPN bracis-28371 164 7 method method NOUN bracis-28371 164 8 to to PART bracis-28371 164 9 handle handle VERB bracis-28371 164 10 dependent dependent ADJ bracis-28371 164 11 features feature NOUN bracis-28371 165 1 [ [ X bracis-28371 165 2 1 1 NUM bracis-28371 165 3 ] ] PUNCT bracis-28371 165 4 , , PUNCT bracis-28371 165 5 etc etc X bracis-28371 165 6 . . X bracis-28371 165 7 ) ) PUNCT bracis-28371 165 8 . . PUNCT bracis-28371 166 1 with with ADP bracis-28371 166 2 the the DET bracis-28371 166 3 shap shap PROPN bracis-28371 166 4 values value NOUN bracis-28371 166 5 , , PUNCT bracis-28371 166 6 we we PRON bracis-28371 166 7 create create VERB bracis-28371 166 8 an an DET bracis-28371 166 9 instance instance NOUN bracis-28371 166 10 \(i'\ \(i'\ NUM bracis-28371 166 11 ) ) PUNCT bracis-28371 166 12 , , PUNCT bracis-28371 166 13 which which PRON bracis-28371 166 14 consists consist VERB bracis-28371 166 15 of of ADP bracis-28371 166 16 the the DET bracis-28371 166 17 calculated calculate VERB bracis-28371 166 18 shap shap PROPN bracis-28371 166 19 values value NOUN bracis-28371 166 20 instead instead ADV bracis-28371 166 21 of of ADP bracis-28371 166 22 the the DET bracis-28371 166 23 raw raw ADJ bracis-28371 166 24 feature feature NOUN bracis-28371 166 25 values value NOUN bracis-28371 166 26 . . PUNCT bracis-28371 167 1 in in ADP bracis-28371 167 2 the the DET bracis-28371 167 3 next next ADJ bracis-28371 167 4 step step NOUN bracis-28371 167 5 ( ( PUNCT bracis-28371 167 6 4.b 4.b NUM bracis-28371 167 7 ) ) PUNCT bracis-28371 167 8 , , PUNCT bracis-28371 167 9 we we PRON bracis-28371 167 10 create create VERB bracis-28371 167 11 a a DET bracis-28371 167 12 dataset dataset NOUN bracis-28371 167 13 c c NOUN bracis-28371 167 14 constituted constitute VERB bracis-28371 167 15 by by ADP bracis-28371 167 16 the the DET bracis-28371 167 17 instances instance NOUN bracis-28371 167 18 \(i'\ \(i'\ NUM bracis-28371 167 19 ) ) PUNCT bracis-28371 167 20 . . PUNCT bracis-28371 168 1 finally finally ADV bracis-28371 168 2 , , PUNCT bracis-28371 168 3 dimensionality dimensionality NOUN bracis-28371 168 4 reduction reduction NOUN bracis-28371 168 5 is be AUX bracis-28371 168 6 applied apply VERB bracis-28371 168 7 to to ADP bracis-28371 168 8 the the DET bracis-28371 168 9 shap shap NOUN bracis-28371 168 10 - - PUNCT bracis-28371 168 11 transformed transform VERB bracis-28371 168 12 instances instance NOUN bracis-28371 168 13 in in ADP bracis-28371 168 14 c c PROPN bracis-28371 168 15 to to PART bracis-28371 168 16 generate generate VERB bracis-28371 168 17 a a DET bracis-28371 168 18 2d 2d NUM bracis-28371 168 19 plot plot NOUN bracis-28371 168 20 that that PRON bracis-28371 168 21 gives give VERB bracis-28371 168 22 insights insight NOUN bracis-28371 168 23 into into ADP bracis-28371 168 24 the the DET bracis-28371 168 25 separability separability NOUN bracis-28371 168 26 of of ADP bracis-28371 168 27 instances instance NOUN bracis-28371 168 28 . . PUNCT bracis-28371 169 1 this this DET bracis-28371 169 2 procedure procedure NOUN bracis-28371 169 3 is be AUX bracis-28371 169 4 represented represent VERB bracis-28371 169 5 by by ADP bracis-28371 169 6 step step NOUN bracis-28371 169 7 5.b 5.b NUM bracis-28371 169 8 . . PUNCT bracis-28371 170 1 we we PRON bracis-28371 170 2 applied apply VERB bracis-28371 170 3 the the DET bracis-28371 170 4 t t PROPN bracis-28371 170 5 - - PUNCT bracis-28371 170 6 sne sne NOUN bracis-28371 170 7 dimensionality dimensionality NOUN bracis-28371 170 8 reduction reduction NOUN bracis-28371 170 9 ( ( PUNCT bracis-28371 170 10 dr dr NOUN bracis-28371 170 11 ) ) PUNCT bracis-28371 170 12 technique technique NOUN bracis-28371 170 13 in in ADP bracis-28371 170 14 our our PRON bracis-28371 170 15 tests test NOUN bracis-28371 170 16 , , PUNCT bracis-28371 170 17 but but CCONJ bracis-28371 170 18 other other ADJ bracis-28371 170 19 dr dr PROPN bracis-28371 170 20 techniques technique NOUN bracis-28371 170 21 could could AUX bracis-28371 170 22 be be AUX bracis-28371 170 23 used use VERB bracis-28371 170 24 . . PUNCT bracis-28371 171 1 the the DET bracis-28371 171 2 representation representation NOUN bracis-28371 171 3 in in ADP bracis-28371 171 4 6.b 6.b NUM bracis-28371 171 5 is be AUX bracis-28371 171 6 the the DET bracis-28371 171 7 result result NOUN bracis-28371 171 8 of of ADP bracis-28371 171 9 the the DET bracis-28371 171 10 projection projection NOUN bracis-28371 171 11 into into ADP bracis-28371 171 12 2d 2d NUM bracis-28371 171 13 generated generate VERB bracis-28371 171 14 by by ADP bracis-28371 171 15 t t PROPN bracis-28371 171 16 - - PUNCT bracis-28371 171 17 sne sne NOUN bracis-28371 171 18 . . PUNCT bracis-28371 172 1 we we PRON bracis-28371 172 2 observe observe VERB bracis-28371 172 3 that that SCONJ bracis-28371 172 4 the the DET bracis-28371 172 5 instances instance NOUN bracis-28371 172 6 may may AUX bracis-28371 172 7 form form VERB bracis-28371 172 8 well well ADV bracis-28371 172 9 - - PUNCT bracis-28371 172 10 defined define VERB bracis-28371 172 11 clusters cluster NOUN bracis-28371 172 12 in in ADP bracis-28371 172 13 the the DET bracis-28371 172 14 projection projection NOUN bracis-28371 172 15 . . PUNCT bracis-28371 173 1 for for ADP bracis-28371 173 2 instance instance NOUN bracis-28371 173 3 , , PUNCT bracis-28371 173 4 certain certain ADJ bracis-28371 173 5 regions region NOUN bracis-28371 173 6 may may AUX bracis-28371 173 7 have have VERB bracis-28371 173 8 a a DET bracis-28371 173 9 prevalence prevalence NOUN bracis-28371 173 10 of of ADP bracis-28371 173 11 a a DET bracis-28371 173 12 specific specific ADJ bracis-28371 173 13 class class NOUN bracis-28371 173 14 , , PUNCT bracis-28371 173 15 while while SCONJ bracis-28371 173 16 others other NOUN bracis-28371 173 17 may may AUX bracis-28371 173 18 show show VERB bracis-28371 173 19 higher high ADJ bracis-28371 173 20 uncertainty uncertainty NOUN bracis-28371 173 21 . . PUNCT bracis-28371 174 1 moreover moreover ADV bracis-28371 174 2 , , PUNCT bracis-28371 174 3 the the DET bracis-28371 174 4 projected project VERB bracis-28371 174 5 points point NOUN bracis-28371 174 6 can can AUX bracis-28371 174 7 be be AUX bracis-28371 174 8 colored color VERB bracis-28371 174 9 based base VERB bracis-28371 174 10 on on ADP bracis-28371 174 11 the the DET bracis-28371 174 12 feature feature NOUN bracis-28371 174 13 values value NOUN bracis-28371 174 14 , , PUNCT bracis-28371 174 15 which which PRON bracis-28371 174 16 makes make VERB bracis-28371 174 17 it it PRON bracis-28371 174 18 possible possible ADJ bracis-28371 174 19 to to PART bracis-28371 174 20 obtain obtain VERB bracis-28371 174 21 important important ADJ bracis-28371 174 22 insights insight NOUN bracis-28371 174 23 , , PUNCT bracis-28371 174 24 such such ADJ bracis-28371 174 25 as as ADP bracis-28371 174 26 the the DET bracis-28371 174 27 relationship relationship NOUN bracis-28371 174 28 between between ADP bracis-28371 174 29 the the DET bracis-28371 174 30 features feature NOUN bracis-28371 174 31 taken take VERB bracis-28371 174 32 into into ADP bracis-28371 174 33 account account NOUN bracis-28371 174 34 by by ADP bracis-28371 174 35 the the DET bracis-28371 174 36 predictive predictive ADJ bracis-28371 174 37 models model NOUN bracis-28371 174 38 and and CCONJ bracis-28371 174 39 their their PRON bracis-28371 174 40 correct correct ADJ bracis-28371 174 41 and and CCONJ bracis-28371 174 42 incorrect incorrect ADJ bracis-28371 174 43 predictions prediction NOUN bracis-28371 174 44 . . PUNCT bracis-28371 175 1 4 4 NUM bracis-28371 175 2 experimental experimental ADJ bracis-28371 175 3 evaluation evaluation NOUN bracis-28371 175 4 in in ADP bracis-28371 175 5 this this DET bracis-28371 175 6 section section NOUN bracis-28371 175 7 , , PUNCT bracis-28371 175 8 we we PRON bracis-28371 175 9 describe describe VERB bracis-28371 175 10 the the DET bracis-28371 175 11 experiments experiment NOUN bracis-28371 175 12 performed perform VERB bracis-28371 175 13 to to PART bracis-28371 175 14 answer answer VERB bracis-28371 175 15 the the DET bracis-28371 175 16 following follow VERB bracis-28371 175 17 research research NOUN bracis-28371 175 18 questions question NOUN bracis-28371 175 19 : : PUNCT bracis-28371 175 20 rq1 rq1 NOUN bracis-28371 175 21 : : PUNCT bracis-28371 175 22 is be AUX bracis-28371 175 23 it it PRON bracis-28371 175 24 possible possible ADJ bracis-28371 175 25 to to PART bracis-28371 175 26 identify identify VERB bracis-28371 175 27 , , PUNCT bracis-28371 175 28 among among ADP bracis-28371 175 29 patients patient NOUN bracis-28371 175 30 with with ADP bracis-28371 175 31 covid-19 covid-19 PROPN bracis-28371 175 32 , , PUNCT bracis-28371 175 33 who who PRON bracis-28371 175 34 is be AUX bracis-28371 175 35 more more ADV bracis-28371 175 36 likely likely ADJ bracis-28371 175 37 to to PART bracis-28371 175 38 die die VERB bracis-28371 175 39 from from ADP bracis-28371 175 40 the the DET bracis-28371 175 41 disease disease NOUN bracis-28371 175 42 ? ? PUNCT bracis-28371 176 1 rq2 rq2 NOUN bracis-28371 176 2 : : PUNCT bracis-28371 176 3 which which DET bracis-28371 176 4 features feature NOUN bracis-28371 176 5 are be AUX bracis-28371 176 6 taken take VERB bracis-28371 176 7 into into ADP bracis-28371 176 8 account account NOUN bracis-28371 176 9 by by ADP bracis-28371 176 10 the the DET bracis-28371 176 11 classifiers classifier NOUN bracis-28371 176 12 for for ADP bracis-28371 176 13 making make VERB bracis-28371 176 14 predictions prediction NOUN bracis-28371 176 15 and and CCONJ bracis-28371 176 16 how how SCONJ bracis-28371 176 17 do do AUX bracis-28371 176 18 they they PRON bracis-28371 176 19 relate relate VERB bracis-28371 176 20 to to PART bracis-28371 176 21 correct correct VERB bracis-28371 176 22 / / SYM bracis-28371 176 23 incorrect incorrect ADJ bracis-28371 176 24 classifications classification NOUN bracis-28371 176 25 ? ? PUNCT bracis-28371 177 1 rq3 rq3 NOUN bracis-28371 177 2 : : PUNCT bracis-28371 177 3 is be AUX bracis-28371 177 4 the the DET bracis-28371 177 5 performance performance NOUN bracis-28371 177 6 obtained obtain VERB bracis-28371 177 7 with with ADP bracis-28371 177 8 multivariable multivariable ADJ bracis-28371 177 9 classifiers classifier NOUN bracis-28371 177 10 superior superior ADJ bracis-28371 177 11 to to ADP bracis-28371 177 12 the the DET bracis-28371 177 13 performance performance NOUN bracis-28371 177 14 obtained obtain VERB bracis-28371 177 15 with with ADP bracis-28371 177 16 single single ADJ bracis-28371 177 17 - - PUNCT bracis-28371 177 18 variable variable ADJ bracis-28371 177 19 classifiers classifier NOUN bracis-28371 177 20 using use VERB bracis-28371 177 21 only only ADJ bracis-28371 177 22 age age NOUN bracis-28371 177 23 as as ADP bracis-28371 177 24 independent independent ADJ bracis-28371 177 25 variable variable NOUN bracis-28371 177 26 ? ? PUNCT bracis-28371 178 1 4.1 4.1 NUM bracis-28371 178 2 experimental experimental ADJ bracis-28371 178 3 setup setup NOUN bracis-28371 178 4 and and CCONJ bracis-28371 178 5     SPACE bracis-28371 178 6 executions execution NOUN bracis-28371 178 7 data datum NOUN bracis-28371 178 8 preprocessing preprocessing NOUN bracis-28371 178 9 was be AUX bracis-28371 178 10 performed perform VERB bracis-28371 178 11 using use VERB bracis-28371 178 12 python python NOUN bracis-28371 178 13 v3.5 v3.5 NOUN bracis-28371 178 14 and and CCONJ bracis-28371 178 15 the the DET bracis-28371 178 16 libraries library NOUN bracis-28371 178 17 pandas panda NOUN bracis-28371 178 18 and and CCONJ bracis-28371 178 19 numpy numpy NOUN bracis-28371 178 20 . . PUNCT bracis-28371 179 1 the the DET bracis-28371 179 2 classifiers classifier NOUN bracis-28371 179 3 were be AUX bracis-28371 179 4 implemented implement VERB bracis-28371 179 5 using use VERB bracis-28371 179 6 the the DET bracis-28371 179 7 weka weka PROPN bracis-28371 179 8 v3.9.6 v3.9.6 ADV bracis-28371 179 9 api api NOUN bracis-28371 179 10 through through ADP bracis-28371 179 11 the the DET bracis-28371 179 12 wrapper wrapper NOUN bracis-28371 179 13 python python NOUN bracis-28371 179 14 - - PUNCT bracis-28371 179 15 weka weka PROPN bracis-28371 179 16 - - PUNCT bracis-28371 179 17 wrapper3 wrapper3 PROPN bracis-28371 179 18 v0.2.9footnote v0.2.9footnote PROPN bracis-28371 180 1 2 2 NUM bracis-28371 180 2 — — PUNCT bracis-28371 180 3 with with ADP bracis-28371 180 4 the the DET bracis-28371 180 5 exception exception NOUN bracis-28371 180 6 of of ADP bracis-28371 180 7 the the DET bracis-28371 180 8 gradient gradient NOUN bracis-28371 180 9 boosting boost VERB bracis-28371 180 10 tree tree NOUN bracis-28371 180 11 , , PUNCT bracis-28371 180 12 for for ADP bracis-28371 180 13 which which PRON bracis-28371 180 14 used use VERB bracis-28371 180 15 the the DET bracis-28371 180 16 xgboost xgboost PROPN bracis-28371 180 17 implementationfootnote implementationfootnote NOUN bracis-28371 180 18 3 3 NUM bracis-28371 180 19 . . PUNCT bracis-28371 180 20 xgboost xgboost PROPN bracis-28371 180 21 can can AUX bracis-28371 180 22 handle handle VERB bracis-28371 180 23 missing miss VERB bracis-28371 180 24 values value NOUN bracis-28371 180 25 , , PUNCT bracis-28371 180 26 so so SCONJ bracis-28371 180 27 no no DET bracis-28371 180 28 data data NOUN bracis-28371 180 29 imputation imputation NOUN bracis-28371 180 30 was be AUX bracis-28371 180 31 needed need VERB bracis-28371 180 32 . . PUNCT bracis-28371 181 1 for for ADP bracis-28371 181 2 the the DET bracis-28371 181 3 other other ADJ bracis-28371 181 4 algorithms algorithm NOUN bracis-28371 181 5 , , PUNCT bracis-28371 181 6 the the DET bracis-28371 181 7 feature feature NOUN bracis-28371 181 8 values value NOUN bracis-28371 181 9 were be AUX bracis-28371 181 10 imputed impute VERB bracis-28371 181 11 with with ADP bracis-28371 181 12 the the DET bracis-28371 181 13 median median ADJ bracis-28371 181 14 value value NOUN bracis-28371 181 15 from from ADP bracis-28371 181 16 the the DET bracis-28371 181 17 training training NOUN bracis-28371 181 18 set set NOUN bracis-28371 181 19 . . PUNCT bracis-28371 182 1 categorical categorical ADJ bracis-28371 182 2 features feature NOUN bracis-28371 182 3 were be AUX bracis-28371 182 4 imputed impute VERB bracis-28371 182 5 with with ADP bracis-28371 182 6 mode mode NOUN bracis-28371 182 7 values value NOUN bracis-28371 182 8 . . PUNCT bracis-28371 183 1 the the DET bracis-28371 183 2 weights weight NOUN bracis-28371 183 3 of of ADP bracis-28371 183 4 the the DET bracis-28371 183 5 cost cost NOUN bracis-28371 183 6 matrix matrix NOUN bracis-28371 183 7 were be AUX bracis-28371 183 8 calculated calculate VERB bracis-28371 183 9 as as SCONJ bracis-28371 183 10 shown show VERB bracis-28371 183 11 in in ADP bracis-28371 183 12 sect sect NOUN bracis-28371 183 13 . . PUNCT bracis-28371 183 14     SPACE bracis-28371 183 15 3.3 3.3 NUM bracis-28371 183 16 , , PUNCT bracis-28371 183 17 considering consider VERB bracis-28371 183 18 the the DET bracis-28371 183 19 imbalance imbalance NOUN bracis-28371 183 20 rate rate NOUN bracis-28371 183 21 of of ADP bracis-28371 183 22 the the DET bracis-28371 183 23 training training NOUN bracis-28371 183 24 set set NOUN bracis-28371 183 25 of of ADP bracis-28371 183 26 each each DET bracis-28371 183 27 bootstrap bootstrap NOUN bracis-28371 183 28 iteration iteration NOUN bracis-28371 183 29 . . PUNCT bracis-28371 184 1 to to PART bracis-28371 184 2 interpret interpret VERB bracis-28371 184 3 the the DET bracis-28371 184 4 generated generate VERB bracis-28371 184 5 models model NOUN bracis-28371 184 6 , , PUNCT bracis-28371 184 7 we we PRON bracis-28371 184 8 use use VERB bracis-28371 184 9 the the DET bracis-28371 184 10 shap shap NOUN bracis-28371 184 11 libraryfootnote libraryfootnote VERB bracis-28371 184 12 4 4 NUM bracis-28371 184 13 . . PUNCT bracis-28371 185 1 shap shap PROPN bracis-28371 185 2 relies rely VERB bracis-28371 185 3 on on ADP bracis-28371 185 4 shapley shapley ADJ bracis-28371 185 5 values value NOUN bracis-28371 185 6 to to PART bracis-28371 185 7 explain explain VERB bracis-28371 185 8 the the DET bracis-28371 185 9 contribution contribution NOUN bracis-28371 185 10 that that PRON bracis-28371 185 11 each each DET bracis-28371 185 12 variable variable NOUN bracis-28371 185 13 makes make VERB bracis-28371 185 14 to to ADP bracis-28371 185 15 the the DET bracis-28371 185 16 output output NOUN bracis-28371 185 17 predicted predict VERB bracis-28371 185 18 by by ADP bracis-28371 185 19 the the DET bracis-28371 185 20 model model NOUN bracis-28371 185 21 . . PUNCT bracis-28371 186 1 specifically specifically ADV bracis-28371 186 2 , , PUNCT bracis-28371 186 3 we we PRON bracis-28371 186 4 used use VERB bracis-28371 186 5 the the DET bracis-28371 186 6 kernelshap kernelshap PROPN bracis-28371 186 7 algorithm algorithm PROPN bracis-28371 186 8 , , PUNCT bracis-28371 186 9 as as SCONJ bracis-28371 186 10 it it PRON bracis-28371 186 11 is be AUX bracis-28371 186 12 model model ADJ bracis-28371 186 13 - - ADJ bracis-28371 186 14 agnostic agnostic ADJ bracis-28371 186 15 . . PUNCT bracis-28371 187 1 the the DET bracis-28371 187 2 perplexity perplexity NOUN bracis-28371 187 3 parameter parameter NOUN bracis-28371 187 4 for for ADP bracis-28371 187 5 the the DET bracis-28371 187 6 t t PROPN bracis-28371 187 7 - - PUNCT bracis-28371 187 8 sne sne PROPN bracis-28371 187 9 projection projection NOUN bracis-28371 187 10 considered consider VERB bracis-28371 187 11 values value NOUN bracis-28371 187 12 between between ADP bracis-28371 187 13 10 10 NUM bracis-28371 187 14 and and CCONJ bracis-28371 187 15 40 40 NUM bracis-28371 187 16 . . PUNCT bracis-28371 188 1 empirical empirical ADJ bracis-28371 188 2 tests test NOUN bracis-28371 188 3 were be AUX bracis-28371 188 4 performed perform VERB bracis-28371 188 5 in in ADP bracis-28371 188 6 multiple multiple ADJ bracis-28371 188 7 executions execution NOUN bracis-28371 188 8 to to PART bracis-28371 188 9 set set VERB bracis-28371 188 10 the the DET bracis-28371 188 11 appropriate appropriate ADJ bracis-28371 188 12 perplexity perplexity NOUN bracis-28371 188 13 value value NOUN bracis-28371 188 14 . . PUNCT bracis-28371 189 1 4.2 4.2 NUM bracis-28371 189 2 results result NOUN bracis-28371 189 3 for for ADP bracis-28371 189 4     SPACE bracis-28371 189 5 mortality mortality NOUN bracis-28371 189 6 prediction prediction NOUN bracis-28371 189 7 ( ( PUNCT bracis-28371 189 8 rq1 rq1 NOUN bracis-28371 189 9 ) ) PUNCT bracis-28371 189 10 table table NOUN bracis-28371 189 11     SPACE bracis-28371 189 12 2 2 NUM bracis-28371 189 13 displays display VERB bracis-28371 189 14 the the DET bracis-28371 189 15 scores score NOUN bracis-28371 189 16 achieved achieve VERB bracis-28371 189 17 by by ADP bracis-28371 189 18 each each DET bracis-28371 189 19 classifier classifier NOUN bracis-28371 189 20 . . PUNCT bracis-28371 190 1 algorithms algorithm NOUN bracis-28371 190 2 have have VERB bracis-28371 190 3 better well ADJ bracis-28371 190 4 / / SYM bracis-28371 190 5 worse bad ADJ bracis-28371 190 6 performances performance NOUN bracis-28371 190 7 across across ADP bracis-28371 190 8 all all DET bracis-28371 190 9 metrics metric NOUN bracis-28371 190 10 , , PUNCT bracis-28371 190 11 with with ADP bracis-28371 190 12 no no DET bracis-28371 190 13 clear clear ADJ bracis-28371 190 14 winner winner NOUN bracis-28371 190 15 . . PUNCT bracis-28371 191 1 in in ADP bracis-28371 191 2 our our PRON bracis-28371 191 3 analysis analysis NOUN bracis-28371 191 4 , , PUNCT bracis-28371 191 5 we we PRON bracis-28371 191 6 give give VERB bracis-28371 191 7 emphasis emphasis NOUN bracis-28371 191 8 to to ADP bracis-28371 191 9 sensitivity sensitivity NOUN bracis-28371 191 10 results result NOUN bracis-28371 191 11 , , PUNCT bracis-28371 191 12 as as SCONJ bracis-28371 191 13 the the DET bracis-28371 191 14 consequences consequence NOUN bracis-28371 191 15 of of ADP bracis-28371 191 16 not not PART bracis-28371 191 17 identifying identify VERB bracis-28371 191 18 a a DET bracis-28371 191 19 patient patient NOUN bracis-28371 191 20 with with ADP bracis-28371 191 21 a a DET bracis-28371 191 22 high high ADJ bracis-28371 191 23 probability probability NOUN bracis-28371 191 24 of of ADP bracis-28371 191 25 dying die VERB bracis-28371 191 26 are be AUX bracis-28371 191 27 much much ADV bracis-28371 191 28 more more ADV bracis-28371 191 29 severe severe ADJ bracis-28371 191 30 than than ADP bracis-28371 191 31 wrongly wrongly ADV bracis-28371 191 32 predicting predict VERB bracis-28371 191 33 a a DET bracis-28371 191 34 patient patient NOUN bracis-28371 191 35 into into ADP bracis-28371 191 36 the the DET bracis-28371 191 37 alive alive ADJ bracis-28371 191 38 class class NOUN bracis-28371 191 39 when when SCONJ bracis-28371 191 40 a a DET bracis-28371 191 41 patient patient NOUN bracis-28371 191 42 has have VERB bracis-28371 191 43 a a DET bracis-28371 191 44 high high ADJ bracis-28371 191 45 probability probability NOUN bracis-28371 191 46 of of ADP bracis-28371 191 47 dying die VERB bracis-28371 191 48 . . PUNCT bracis-28371 192 1 considering consider VERB bracis-28371 192 2 only only ADJ bracis-28371 192 3 sensitivity sensitivity NOUN bracis-28371 192 4 , , PUNCT bracis-28371 192 5 logistic logistic ADJ bracis-28371 192 6 regression regression NOUN bracis-28371 192 7 is be AUX bracis-28371 192 8 the the DET bracis-28371 192 9 winner winner NOUN bracis-28371 192 10 , , PUNCT bracis-28371 192 11 leading lead VERB bracis-28371 192 12 all all DET bracis-28371 192 13 scenarios scenario NOUN bracis-28371 192 14 . . PUNCT bracis-28371 193 1 only only ADV bracis-28371 193 2 in in ADP bracis-28371 193 3     SPACE bracis-28371 193 4 \(\text \(\text NOUN bracis-28371 193 5 { { PUNCT bracis-28371 193 6 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 193 7 ) ) PUNCT bracis-28371 193 8 xgboost xgboost X bracis-28371 193 9 reaches reach VERB bracis-28371 193 10 the the DET bracis-28371 193 11 same same ADJ bracis-28371 193 12 top top ADJ bracis-28371 193 13 score score NOUN bracis-28371 193 14 as as ADP bracis-28371 193 15 lr lr PROPN bracis-28371 193 16 . . PUNCT bracis-28371 194 1 adaboostm1 adaboostm1 PROPN bracis-28371 194 2 has have VERB bracis-28371 194 3 the the DET bracis-28371 194 4 second second ADJ bracis-28371 194 5 highest high ADJ bracis-28371 194 6 sensitivity sensitivity NOUN bracis-28371 194 7 values value NOUN bracis-28371 194 8 for for ADP bracis-28371 194 9 \(\text \(\text NOUN bracis-28371 194 10 { { PUNCT bracis-28371 194 11 hmv}_{all}\ hmv}_{all}\ PROPN bracis-28371 194 12 ) ) PUNCT bracis-28371 194 13 and and CCONJ bracis-28371 194 14 \(\text \(\text NOUN bracis-28371 194 15 { { PUNCT bracis-28371 194 16 hmv}_{cti}\ hmv}_{cti}\ PROPN bracis-28371 194 17 ) ) PUNCT bracis-28371 194 18 , , PUNCT bracis-28371 194 19 and and CCONJ bracis-28371 194 20 the the DET bracis-28371 194 21 third third ADJ bracis-28371 194 22 highest high ADJ bracis-28371 194 23 sensitivity sensitivity NOUN bracis-28371 194 24 value value NOUN bracis-28371 194 25 for for ADP bracis-28371 194 26 \(\text \(\text PROPN bracis-28371 194 27 { { PUNCT bracis-28371 194 28 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 194 29 ) ) PUNCT bracis-28371 194 30 . . PUNCT bracis-28371 195 1 in in ADP bracis-28371 195 2 all all DET bracis-28371 195 3 cases case NOUN bracis-28371 195 4 , , PUNCT bracis-28371 195 5 j48 j48 PROPN bracis-28371 195 6 has have VERB bracis-28371 195 7 the the DET bracis-28371 195 8 lowest low ADJ bracis-28371 195 9 sensitivity sensitivity NOUN bracis-28371 195 10 values value NOUN bracis-28371 195 11 . . PUNCT bracis-28371 196 1 table table NOUN bracis-28371 196 2 2 2 NUM bracis-28371 196 3 . . X bracis-28371 196 4 results result NOUN bracis-28371 196 5 for for ADP bracis-28371 196 6 mortality mortality NOUN bracis-28371 196 7 predictionfull predictionfull NOUN bracis-28371 196 8 size size NOUN bracis-28371 196 9 table table NOUN bracis-28371 196 10 overall overall ADV bracis-28371 196 11 , , PUNCT bracis-28371 196 12 the the DET bracis-28371 196 13 sensitivity sensitivity NOUN bracis-28371 196 14 values value NOUN bracis-28371 196 15 we we PRON bracis-28371 196 16 obtain obtain VERB bracis-28371 196 17 demonstrate demonstrate VERB bracis-28371 196 18 the the DET bracis-28371 196 19 classifier classifier NOUN bracis-28371 196 20 ’s ’s PART bracis-28371 196 21 promising promising ADJ bracis-28371 196 22 ability ability NOUN bracis-28371 196 23 to to PART bracis-28371 196 24 predict predict VERB bracis-28371 196 25 mortality mortality NOUN bracis-28371 196 26 based base VERB bracis-28371 196 27 on on ADP bracis-28371 196 28 the the DET bracis-28371 196 29 information information NOUN bracis-28371 196 30 available available ADJ bracis-28371 196 31 at at ADP bracis-28371 196 32 the the DET bracis-28371 196 33 time time NOUN bracis-28371 196 34 of of ADP bracis-28371 196 35 admission admission NOUN bracis-28371 196 36 ( ( PUNCT bracis-28371 196 37 answer answer NOUN bracis-28371 196 38 to to PART bracis-28371 196 39 rq1 rq1 VERB bracis-28371 196 40 ) ) PUNCT bracis-28371 196 41 . . PUNCT bracis-28371 197 1 the the DET bracis-28371 197 2 average average ADJ bracis-28371 197 3 sensitivity sensitivity NOUN bracis-28371 197 4 for for ADP bracis-28371 197 5 the the DET bracis-28371 197 6 classification classification NOUN bracis-28371 197 7 algorithms algorithms NOUN bracis-28371 197 8 is be AUX bracis-28371 197 9 higher high ADJ bracis-28371 197 10 for for ADP bracis-28371 197 11 the the DET bracis-28371 197 12 \(\text \(\text NOUN bracis-28371 197 13 { { PUNCT bracis-28371 197 14 hmv}_{cti}\ hmv}_{cti}\ PROPN bracis-28371 197 15 ) ) PUNCT bracis-28371 197 16 dataset dataset NOUN bracis-28371 197 17 ( ( PUNCT bracis-28371 197 18 0.74 0.74 NUM bracis-28371 197 19 ) ) PUNCT bracis-28371 197 20 when when SCONJ bracis-28371 197 21 compared compare VERB bracis-28371 197 22 to to ADP bracis-28371 197 23 the the DET bracis-28371 197 24 \(\text \(\text NOUN bracis-28371 197 25 { { PUNCT bracis-28371 197 26 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 197 27 ) ) PUNCT bracis-28371 197 28 dataset dataset NOUN bracis-28371 197 29 ( ( PUNCT bracis-28371 197 30 0.65 0.65 NUM bracis-28371 197 31 ) ) PUNCT bracis-28371 197 32 . . PUNCT bracis-28371 198 1 this this DET bracis-28371 198 2 difference difference NOUN bracis-28371 198 3 may may AUX bracis-28371 198 4 be be AUX bracis-28371 198 5 explained explain VERB bracis-28371 198 6 by by ADP bracis-28371 198 7 the the DET bracis-28371 198 8 individual individual ADJ bracis-28371 198 9 characteristics characteristic NOUN bracis-28371 198 10 of of ADP bracis-28371 198 11 patients patient NOUN bracis-28371 198 12 admitted admit VERB bracis-28371 198 13 to to ADP bracis-28371 198 14 each each DET bracis-28371 198 15 hospital hospital NOUN bracis-28371 198 16 and and CCONJ bracis-28371 198 17 the the DET bracis-28371 198 18 features feature NOUN bracis-28371 198 19 available available ADJ bracis-28371 198 20 in in ADP bracis-28371 198 21 the the DET bracis-28371 198 22 datasets dataset NOUN bracis-28371 198 23 for for ADP bracis-28371 198 24 each each DET bracis-28371 198 25 hospital hospital NOUN bracis-28371 198 26 . . PUNCT bracis-28371 199 1 a a DET bracis-28371 199 2 similar similar ADJ bracis-28371 199 3 discussion discussion NOUN bracis-28371 199 4 is be AUX bracis-28371 199 5 mentioned mention VERB bracis-28371 199 6 in in ADP bracis-28371 199 7 futoma futoma NOUN bracis-28371 199 8     SPACE bracis-28371 199 9 et et PROPN bracis-28371 199 10 al al PROPN bracis-28371 199 11 . . PUNCT bracis-28371 199 12     SPACE bracis-28371 200 1 [ [ X bracis-28371 200 2 6 6 NUM bracis-28371 200 3 ] ] PUNCT bracis-28371 200 4 , , PUNCT bracis-28371 200 5 which which PRON bracis-28371 200 6 warns warn VERB bracis-28371 200 7 about about ADP bracis-28371 200 8 the the DET bracis-28371 200 9 risks risk NOUN bracis-28371 200 10 of of ADP bracis-28371 200 11 excessively excessively ADV bracis-28371 200 12 seeking seek VERB bracis-28371 200 13 the the DET bracis-28371 200 14 development development NOUN bracis-28371 200 15 of of ADP bracis-28371 200 16 predictive predictive ADJ bracis-28371 200 17 models model NOUN bracis-28371 200 18 that that PRON bracis-28371 200 19 generalize generalize VERB bracis-28371 200 20 to to ADP bracis-28371 200 21 various various ADJ bracis-28371 200 22 hospitals hospital NOUN bracis-28371 200 23 with with ADP bracis-28371 200 24 different different ADJ bracis-28371 200 25 clinical clinical ADJ bracis-28371 200 26 protocols protocol NOUN bracis-28371 200 27 and and CCONJ bracis-28371 200 28 units unit NOUN bracis-28371 200 29 . . PUNCT bracis-28371 201 1 our our PRON bracis-28371 201 2 best good ADJ bracis-28371 201 3 auroc auroc NOUN bracis-28371 201 4 is be AUX bracis-28371 201 5 higher high ADJ bracis-28371 201 6 than than ADP bracis-28371 201 7 the the DET bracis-28371 201 8 ones one NOUN bracis-28371 201 9 reported report VERB bracis-28371 201 10 in in ADP bracis-28371 201 11 related related ADJ bracis-28371 201 12 work work NOUN bracis-28371 201 13     SPACE bracis-28371 202 1 [ [ X bracis-28371 202 2 5 5 NUM bracis-28371 202 3 , , PUNCT bracis-28371 202 4 9 9 NUM bracis-28371 202 5 , , PUNCT bracis-28371 202 6 13 13 NUM bracis-28371 202 7 , , PUNCT bracis-28371 202 8 20 20 NUM bracis-28371 202 9 ] ] PUNCT bracis-28371 202 10 ( ( PUNCT bracis-28371 202 11 see see VERB bracis-28371 202 12 table table NOUN bracis-28371 202 13     SPACE bracis-28371 202 14 2 2 NUM bracis-28371 202 15 ) ) PUNCT bracis-28371 202 16 . . PUNCT bracis-28371 203 1 yet yet CCONJ bracis-28371 203 2 a a DET bracis-28371 203 3 direct direct ADJ bracis-28371 203 4 comparison comparison NOUN bracis-28371 203 5 among among ADP bracis-28371 203 6 approaches approach NOUN bracis-28371 203 7 is be AUX bracis-28371 203 8 not not PART bracis-28371 203 9 possible possible ADJ bracis-28371 203 10 as as SCONJ bracis-28371 203 11 they they PRON bracis-28371 203 12 were be AUX bracis-28371 203 13 applied apply VERB bracis-28371 203 14 to to ADP bracis-28371 203 15 different different ADJ bracis-28371 203 16 datasets dataset NOUN bracis-28371 203 17 . . PUNCT bracis-28371 204 1 4.3 4.3 NUM bracis-28371 204 2 analysis analysis NOUN bracis-28371 204 3 of of ADP bracis-28371 204 4     SPACE bracis-28371 204 5 feature feature NOUN bracis-28371 204 6 importance importance NOUN bracis-28371 204 7 and and CCONJ bracis-28371 204 8     SPACE bracis-28371 204 9 explainability explainability NOUN bracis-28371 204 10 ( ( PUNCT bracis-28371 204 11 rq2 rq2 NOUN bracis-28371 204 12 ) ) PUNCT bracis-28371 204 13 we we PRON bracis-28371 204 14 use use VERB bracis-28371 204 15 the the DET bracis-28371 204 16 results result NOUN bracis-28371 204 17 of of ADP bracis-28371 204 18 applying apply VERB bracis-28371 204 19 dimensionality dimensionality NOUN bracis-28371 204 20 reduction reduction NOUN bracis-28371 204 21 over over ADP bracis-28371 204 22 shap shap PROPN bracis-28371 204 23 values value NOUN bracis-28371 204 24 to to PART bracis-28371 204 25 help help VERB bracis-28371 204 26 understand understand VERB bracis-28371 204 27 the the DET bracis-28371 204 28 factors factor NOUN bracis-28371 204 29 considered consider VERB bracis-28371 204 30 by by ADP bracis-28371 204 31 classifiers classifier NOUN bracis-28371 204 32 during during ADP bracis-28371 204 33 predictions prediction NOUN bracis-28371 204 34 and and CCONJ bracis-28371 204 35 answer answer NOUN bracis-28371 204 36 rq2 rq2 PROPN bracis-28371 204 37 . . PUNCT bracis-28371 205 1 figure figure NOUN bracis-28371 205 2     SPACE bracis-28371 205 3 3 3 NUM bracis-28371 205 4 presents present VERB bracis-28371 205 5 the the DET bracis-28371 205 6 global global ADJ bracis-28371 205 7 view view NOUN bracis-28371 205 8 of of ADP bracis-28371 205 9 how how SCONJ bracis-28371 205 10 features feature NOUN bracis-28371 205 11 contribute contribute VERB bracis-28371 205 12 to to ADP bracis-28371 205 13 mortality mortality NOUN bracis-28371 205 14 prediction prediction NOUN bracis-28371 205 15 using use VERB bracis-28371 205 16 the the DET bracis-28371 205 17 lr lr NOUN bracis-28371 205 18 classifier classifier NOUN bracis-28371 205 19 on on ADP bracis-28371 205 20 \(\text \(\text PROPN bracis-28371 205 21 { { PUNCT bracis-28371 205 22 hmv}_{all}\ hmv}_{all}\ PROPN bracis-28371 205 23 ) ) PUNCT bracis-28371 205 24 . . PUNCT bracis-28371 206 1 each each DET bracis-28371 206 2 line line NOUN bracis-28371 206 3 refers refer VERB bracis-28371 206 4 to to ADP bracis-28371 206 5 the the DET bracis-28371 206 6 contribution contribution NOUN bracis-28371 206 7 of of ADP bracis-28371 206 8 a a DET bracis-28371 206 9 feature feature NOUN bracis-28371 206 10 . . PUNCT bracis-28371 207 1 each each DET bracis-28371 207 2 dot dot NOUN bracis-28371 207 3 is be AUX bracis-28371 207 4 an an DET bracis-28371 207 5 instance instance NOUN bracis-28371 207 6 . . PUNCT bracis-28371 208 1 low low ADJ bracis-28371 208 2 values value NOUN bracis-28371 208 3 for for ADP bracis-28371 208 4 the the DET bracis-28371 208 5 feature feature NOUN bracis-28371 208 6 are be AUX bracis-28371 208 7 in in ADP bracis-28371 208 8 blue blue ADJ bracis-28371 208 9 and and CCONJ bracis-28371 208 10 high high ADJ bracis-28371 208 11 values value NOUN bracis-28371 208 12 are be AUX bracis-28371 208 13 in in ADP bracis-28371 208 14 red red ADJ bracis-28371 208 15 . . PUNCT bracis-28371 209 1 the the DET bracis-28371 209 2 contribution contribution NOUN bracis-28371 209 3 of of ADP bracis-28371 209 4 the the DET bracis-28371 209 5 feature feature NOUN bracis-28371 209 6 value value NOUN bracis-28371 209 7 is be AUX bracis-28371 209 8 given give VERB bracis-28371 209 9 by by ADP bracis-28371 209 10 the the DET bracis-28371 209 11 horizontal horizontal ADJ bracis-28371 209 12 displacement displacement NOUN bracis-28371 209 13 . . PUNCT bracis-28371 210 1 the the DET bracis-28371 210 2 further further ADJ bracis-28371 210 3 the the DET bracis-28371 210 4 instance instance NOUN bracis-28371 210 5 is be AUX bracis-28371 210 6 positioned position VERB bracis-28371 210 7 to to ADP bracis-28371 210 8 the the DET bracis-28371 210 9 right right NOUN bracis-28371 210 10 , , PUNCT bracis-28371 210 11 the the PRON bracis-28371 210 12 greater great ADJ bracis-28371 210 13 the the DET bracis-28371 210 14 contribution contribution NOUN bracis-28371 210 15 to to ADP bracis-28371 210 16 the the DET bracis-28371 210 17 deceased deceased ADJ bracis-28371 210 18 class class NOUN bracis-28371 210 19 . . PUNCT bracis-28371 211 1 the the DET bracis-28371 211 2 further further ADJ bracis-28371 211 3 to to ADP bracis-28371 211 4 the the DET bracis-28371 211 5 left left NOUN bracis-28371 211 6 , , PUNCT bracis-28371 211 7 the the PRON bracis-28371 211 8 greater great ADJ bracis-28371 211 9 the the DET bracis-28371 211 10 contribution contribution NOUN bracis-28371 211 11 of of ADP bracis-28371 211 12 the the DET bracis-28371 211 13 feature feature NOUN bracis-28371 211 14 value value NOUN bracis-28371 211 15 to to ADP bracis-28371 211 16 the the DET bracis-28371 211 17 prediction prediction NOUN bracis-28371 211 18 of of ADP bracis-28371 211 19 the the DET bracis-28371 211 20 alive alive ADJ bracis-28371 211 21 class class NOUN bracis-28371 211 22 . . PUNCT bracis-28371 212 1 for for ADP bracis-28371 212 2 example example NOUN bracis-28371 212 3 , , PUNCT bracis-28371 212 4 high high ADJ bracis-28371 212 5 values value NOUN bracis-28371 212 6 for for ADP bracis-28371 212 7 age age NOUN bracis-28371 212 8 are be AUX bracis-28371 212 9 associated associate VERB bracis-28371 212 10 with with ADP bracis-28371 212 11 increased increase VERB bracis-28371 212 12 probability probability NOUN bracis-28371 212 13 for for ADP bracis-28371 212 14 the the DET bracis-28371 212 15 deceased deceased ADJ bracis-28371 212 16 class class NOUN bracis-28371 212 17 . . PUNCT bracis-28371 213 1 the the DET bracis-28371 213 2 ten ten NUM bracis-28371 213 3 features feature NOUN bracis-28371 213 4 with with ADP bracis-28371 213 5 the the DET bracis-28371 213 6 highest high ADJ bracis-28371 213 7 contribution contribution NOUN bracis-28371 213 8 to to ADP bracis-28371 213 9 the the DET bracis-28371 213 10 predictions prediction NOUN bracis-28371 213 11 were be AUX bracis-28371 213 12 age age NOUN bracis-28371 213 13 , , PUNCT bracis-28371 213 14 erythrocytes erythrocyte NOUN bracis-28371 213 15 , , PUNCT bracis-28371 213 16 bilirubin bilirubin NOUN bracis-28371 213 17 , , PUNCT bracis-28371 213 18 oxygen oxygen NOUN bracis-28371 213 19 saturation saturation NOUN bracis-28371 213 20 , , PUNCT bracis-28371 213 21 neutrophils neutrophil NOUN bracis-28371 213 22 , , PUNCT bracis-28371 213 23 troponin troponin ADJ bracis-28371 213 24 , , PUNCT bracis-28371 213 25 heart heart NOUN bracis-28371 213 26 disease disease NOUN bracis-28371 213 27 , , PUNCT bracis-28371 213 28 hemoglobin hemoglobin NOUN bracis-28371 213 29 , , PUNCT bracis-28371 213 30 creatinine creatinine NOUN bracis-28371 213 31 , , PUNCT bracis-28371 213 32 and and CCONJ bracis-28371 213 33 central central ADJ bracis-28371 213 34 nervous nervous ADJ bracis-28371 213 35 system system NOUN bracis-28371 213 36 disease disease NOUN bracis-28371 213 37 . . PUNCT bracis-28371 214 1 using use VERB bracis-28371 214 2 the the DET bracis-28371 214 3 lr lr NOUN bracis-28371 214 4 classifier classifier NOUN bracis-28371 214 5 on on ADP bracis-28371 214 6 \(\text \(\text PROPN bracis-28371 214 7 { { PUNCT bracis-28371 214 8 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 214 9 ) ) PUNCT bracis-28371 214 10 , , PUNCT bracis-28371 214 11 the the DET bracis-28371 214 12 ten ten NUM bracis-28371 214 13 features feature NOUN bracis-28371 214 14 with with ADP bracis-28371 214 15 the the DET bracis-28371 214 16 highest high ADJ bracis-28371 214 17 contribution contribution NOUN bracis-28371 214 18 were be AUX bracis-28371 214 19 age age NOUN bracis-28371 214 20 , , PUNCT bracis-28371 214 21 urea urea NOUN bracis-28371 214 22 , , PUNCT bracis-28371 214 23 red red ADJ bracis-28371 214 24 cell cell NOUN bracis-28371 214 25 distribution distribution NOUN bracis-28371 214 26 width width NOUN bracis-28371 214 27 , , PUNCT bracis-28371 214 28 d d NOUN bracis-28371 214 29 - - PUNCT bracis-28371 214 30 dimer dimer ADJ bracis-28371 214 31 , , PUNCT bracis-28371 214 32 mean mean ADJ bracis-28371 214 33 corpuscular corpuscular NOUN bracis-28371 214 34 volume volume NOUN bracis-28371 214 35 ( ( PUNCT bracis-28371 214 36 mcv mcv PROPN bracis-28371 214 37 ) ) PUNCT bracis-28371 214 38 , , PUNCT bracis-28371 214 39 lymphocytes lymphocyte NOUN bracis-28371 214 40 , , PUNCT bracis-28371 214 41 platelets platelet NOUN bracis-28371 214 42 , , PUNCT bracis-28371 214 43 creatinine creatinine NOUN bracis-28371 214 44 , , PUNCT bracis-28371 214 45 erythrocytes erythrocyte NOUN bracis-28371 214 46 , , PUNCT bracis-28371 214 47 and and CCONJ bracis-28371 214 48 troponin troponin PROPN bracis-28371 215 1 . . PUNCT bracis-28371 215 2 figure figure NOUN bracis-28371 215 3     SPACE bracis-28371 215 4 4 4 NUM bracis-28371 215 5 shows show VERB bracis-28371 215 6 the the DET bracis-28371 215 7 t t PROPN bracis-28371 215 8 - - PUNCT bracis-28371 215 9 sne sne NOUN bracis-28371 215 10 plots plot NOUN bracis-28371 215 11 generated generate VERB bracis-28371 215 12 using use VERB bracis-28371 215 13 shap shap NOUN bracis-28371 215 14 values value NOUN bracis-28371 215 15 . . PUNCT bracis-28371 216 1 each each DET bracis-28371 216 2 point point NOUN bracis-28371 216 3 corresponds correspond VERB bracis-28371 216 4 to to ADP bracis-28371 216 5 a a DET bracis-28371 216 6 patient patient NOUN bracis-28371 216 7 . . PUNCT bracis-28371 217 1 the the DET bracis-28371 217 2 points point NOUN bracis-28371 217 3 are be AUX bracis-28371 217 4 colored color VERB bracis-28371 217 5 based base VERB bracis-28371 217 6 on on ADP bracis-28371 217 7 the the DET bracis-28371 217 8 values value NOUN bracis-28371 217 9 of of ADP bracis-28371 217 10 different different ADJ bracis-28371 217 11 features feature NOUN bracis-28371 217 12 to to PART bracis-28371 217 13 aid aid VERB bracis-28371 217 14 in in ADP bracis-28371 217 15 interpreting interpret VERB bracis-28371 217 16 the the DET bracis-28371 217 17 generated generate VERB bracis-28371 217 18 projections projection NOUN bracis-28371 217 19 . . PUNCT bracis-28371 218 1 each each DET bracis-28371 218 2 row row NOUN bracis-28371 218 3 is be AUX bracis-28371 218 4 composed compose VERB bracis-28371 218 5 of of ADP bracis-28371 218 6 three three NUM bracis-28371 218 7 plots plot NOUN bracis-28371 218 8 . . PUNCT bracis-28371 219 1 the the DET bracis-28371 219 2 first first ADJ bracis-28371 219 3 plot plot NOUN bracis-28371 219 4 displays display VERB bracis-28371 219 5 the the DET bracis-28371 219 6 correct correct ADJ bracis-28371 219 7 and and CCONJ bracis-28371 219 8 incorrect incorrect ADJ bracis-28371 219 9 predictions prediction NOUN bracis-28371 219 10 ( ( PUNCT bracis-28371 219 11 true true ADJ bracis-28371 219 12 positive positive ADJ bracis-28371 219 13 ( ( PUNCT bracis-28371 219 14 tp tp NOUN bracis-28371 219 15 ) ) PUNCT bracis-28371 219 16 , , PUNCT bracis-28371 219 17 true true ADJ bracis-28371 219 18 negative negative ADJ bracis-28371 219 19 ( ( PUNCT bracis-28371 219 20 tn tn NOUN bracis-28371 219 21 ) ) PUNCT bracis-28371 219 22 , , PUNCT bracis-28371 219 23 false false ADJ bracis-28371 219 24 positive positive ADJ bracis-28371 219 25 ( ( PUNCT bracis-28371 219 26 fp fp NOUN bracis-28371 219 27 ) ) PUNCT bracis-28371 219 28 and and CCONJ bracis-28371 219 29 false false ADJ bracis-28371 219 30 negative(fn negative(fn NOUN bracis-28371 219 31 ) ) PUNCT bracis-28371 219 32 ) ) PUNCT bracis-28371 219 33 . . PUNCT bracis-28371 220 1 the the DET bracis-28371 220 2 second second ADJ bracis-28371 220 3 and and CCONJ bracis-28371 220 4 third third ADJ bracis-28371 220 5 plots plot NOUN bracis-28371 220 6 show show VERB bracis-28371 220 7 points point NOUN bracis-28371 220 8 colored color VERB bracis-28371 220 9 by by ADP bracis-28371 220 10 different different ADJ bracis-28371 220 11 features feature NOUN bracis-28371 220 12 . . PUNCT bracis-28371 221 1 fig fig NOUN bracis-28371 221 2 . . PUNCT bracis-28371 222 1 3 3 X bracis-28371 222 2 . . X bracis-28371 222 3 global global ADJ bracis-28371 222 4 explanations explanation NOUN bracis-28371 222 5 based base VERB bracis-28371 222 6 on on ADP bracis-28371 222 7 lr lr NOUN bracis-28371 222 8 classifier classifier NOUN bracis-28371 222 9 using use VERB bracis-28371 222 10 shap shap NOUN bracis-28371 222 11 for for ADP bracis-28371 222 12     SPACE bracis-28371 222 13 \(\text \(\text PROPN bracis-28371 222 14 { { PUNCT bracis-28371 222 15 hmv}_{all}\ hmv}_{all}\ PROPN bracis-28371 222 16 ) ) PUNCT bracis-28371 222 17 full full ADJ bracis-28371 222 18 size size NOUN bracis-28371 222 19 image image NOUN bracis-28371 222 20 fig fig NOUN bracis-28371 222 21 . . PUNCT bracis-28371 223 1 4 4 X bracis-28371 223 2 . . NOUN bracis-28371 223 3 visualizations visualization NOUN bracis-28371 223 4 of of ADP bracis-28371 223 5 the the DET bracis-28371 223 6 model model NOUN bracis-28371 223 7 ’s ’s PART bracis-28371 223 8 predictions prediction NOUN bracis-28371 223 9 based base VERB bracis-28371 223 10 on on ADP bracis-28371 223 11 shap shap PROPN bracis-28371 223 12 values value NOUN bracis-28371 223 13 – – PUNCT bracis-28371 223 14 the the DET bracis-28371 223 15 x x ADJ bracis-28371 223 16 - - ADJ bracis-28371 223 17 axis axis ADJ bracis-28371 223 18 corresponds correspond NOUN bracis-28371 223 19 to to ADP bracis-28371 223 20 t t PROPN bracis-28371 223 21 - - PUNCT bracis-28371 223 22 sne sne NOUN bracis-28371 223 23 dimension dimension NOUN bracis-28371 223 24 1 1 NUM bracis-28371 223 25 and and CCONJ bracis-28371 223 26 the the DET bracis-28371 223 27 y y NOUN bracis-28371 223 28 - - PUNCT bracis-28371 223 29 axis axis NOUN bracis-28371 223 30 to to ADP bracis-28371 223 31 t t PROPN bracis-28371 223 32 - - PUNCT bracis-28371 223 33 sne sne NOUN bracis-28371 223 34 dimension dimension NOUN bracis-28371 223 35 2 2 NUM bracis-28371 223 36 . . PUNCT bracis-28371 223 37 full full ADJ bracis-28371 223 38 size size NOUN bracis-28371 223 39 image image NOUN bracis-28371 223 40 the the DET bracis-28371 223 41 t t PROPN bracis-28371 223 42 - - PUNCT bracis-28371 223 43 sne sne NOUN bracis-28371 223 44 plots plot NOUN bracis-28371 223 45 reinforce reinforce VERB bracis-28371 223 46 the the DET bracis-28371 223 47 high high ADJ bracis-28371 223 48 importance importance NOUN bracis-28371 223 49 attributed attribute VERB bracis-28371 223 50 by by ADP bracis-28371 223 51 the the DET bracis-28371 223 52 predictive predictive ADJ bracis-28371 223 53 models model NOUN bracis-28371 223 54 to to ADP bracis-28371 223 55 the the DET bracis-28371 223 56 age age NOUN bracis-28371 223 57 feature feature NOUN bracis-28371 223 58 . . PUNCT bracis-28371 224 1 when when SCONJ bracis-28371 224 2 analyzing analyze VERB bracis-28371 224 3 the the DET bracis-28371 224 4 plots plot NOUN bracis-28371 224 5 with with ADP bracis-28371 224 6 the the DET bracis-28371 224 7 classifications classification NOUN bracis-28371 224 8 and and CCONJ bracis-28371 224 9 the the DET bracis-28371 224 10 colored colored ADJ bracis-28371 224 11 plots plot NOUN bracis-28371 224 12 based base VERB bracis-28371 224 13 on on ADP bracis-28371 224 14 the the DET bracis-28371 224 15 age age NOUN bracis-28371 224 16 of of ADP bracis-28371 224 17 the the DET bracis-28371 224 18 patients patient NOUN bracis-28371 224 19 ( ( PUNCT bracis-28371 224 20 figs fig NOUN bracis-28371 224 21 . . PUNCT bracis-28371 224 22     SPACE bracis-28371 224 23 4b 4b PROPN bracis-28371 224 24 , , PUNCT bracis-28371 224 25     SPACE bracis-28371 224 26 4e 4e PROPN bracis-28371 224 27     SPACE bracis-28371 224 28 4h 4h NOUN bracis-28371 224 29 , , PUNCT bracis-28371 224 30 and and CCONJ bracis-28371 224 31     SPACE bracis-28371 224 32 4k 4k NOUN bracis-28371 224 33 ) ) PUNCT bracis-28371 224 34 , , PUNCT bracis-28371 224 35 it it PRON bracis-28371 224 36 is be AUX bracis-28371 224 37 possible possible ADJ bracis-28371 224 38 to to PART bracis-28371 224 39 observe observe VERB bracis-28371 224 40 that that DET bracis-28371 224 41 recurrent recurrent NOUN bracis-28371 224 42 fp fp NOUN bracis-28371 224 43 errors error NOUN bracis-28371 224 44 ( ( PUNCT bracis-28371 224 45 purple purple ADJ bracis-28371 224 46 dots dot NOUN bracis-28371 224 47 ) ) PUNCT bracis-28371 224 48 and and CCONJ bracis-28371 224 49 tp tp NOUN bracis-28371 224 50 occur occur VERB bracis-28371 224 51 in in ADP bracis-28371 224 52 the the DET bracis-28371 224 53 regions region NOUN bracis-28371 224 54 with with ADP bracis-28371 224 55 a a DET bracis-28371 224 56 high high ADJ bracis-28371 224 57 concentration concentration NOUN bracis-28371 224 58 of of ADP bracis-28371 224 59 older old ADJ bracis-28371 224 60 patients patient NOUN bracis-28371 224 61 . . PUNCT bracis-28371 225 1 on on ADP bracis-28371 225 2 the the DET bracis-28371 225 3 other other ADJ bracis-28371 225 4 hand hand NOUN bracis-28371 225 5 , , PUNCT bracis-28371 225 6 fn fn NOUN bracis-28371 225 7 errors error NOUN bracis-28371 225 8 ( ( PUNCT bracis-28371 225 9 red red ADJ bracis-28371 225 10 dots dot NOUN bracis-28371 225 11 ) ) PUNCT bracis-28371 225 12 and and CCONJ bracis-28371 225 13 tn tn NOUN bracis-28371 225 14 are be AUX bracis-28371 225 15 in in ADP bracis-28371 225 16 regions region NOUN bracis-28371 225 17 with with ADP bracis-28371 225 18 younger young ADJ bracis-28371 225 19 patients patient NOUN bracis-28371 225 20 . . PUNCT bracis-28371 226 1 figure figure VERB bracis-28371 226 2     SPACE bracis-28371 226 3 4c 4c NOUN bracis-28371 226 4 also also ADV bracis-28371 226 5 shows show VERB bracis-28371 226 6 a a DET bracis-28371 226 7 trend trend NOUN bracis-28371 226 8 of of ADP bracis-28371 226 9 moderate moderate ADJ bracis-28371 226 10 and and CCONJ bracis-28371 226 11 low low ADJ bracis-28371 226 12 erythrocyte erythrocyte NOUN bracis-28371 226 13 values value NOUN bracis-28371 226 14 for for ADP bracis-28371 226 15 tp tp NOUN bracis-28371 226 16 and and CCONJ bracis-28371 226 17 fp fp ADJ bracis-28371 226 18 patients patient NOUN bracis-28371 226 19 . . PUNCT bracis-28371 227 1 these these DET bracis-28371 227 2 findings finding NOUN bracis-28371 227 3 are be AUX bracis-28371 227 4 supported support VERB bracis-28371 227 5 by by ADP bracis-28371 227 6 previous previous ADJ bracis-28371 227 7 studies study NOUN bracis-28371 227 8 , , PUNCT bracis-28371 227 9 which which PRON bracis-28371 227 10 showed show VERB bracis-28371 227 11 that that SCONJ bracis-28371 227 12 anemia anemia NOUN bracis-28371 227 13 is be AUX bracis-28371 227 14 associated associate VERB bracis-28371 227 15 with with ADP bracis-28371 227 16 poor poor ADJ bracis-28371 227 17 prognosis prognosis NOUN bracis-28371 227 18 in in ADP bracis-28371 227 19 critically critically ADV bracis-28371 227 20 ill ill ADJ bracis-28371 227 21 patients patient NOUN bracis-28371 227 22 . . PUNCT bracis-28371 228 1 figures figure NOUN bracis-28371 228 2     SPACE bracis-28371 228 3 4d 4d NUM bracis-28371 228 4 , , PUNCT bracis-28371 228 5     SPACE bracis-28371 228 6 4e 4e NOUN bracis-28371 228 7 , , PUNCT bracis-28371 228 8 and and CCONJ bracis-28371 228 9     SPACE bracis-28371 228 10 4f 4f NUM bracis-28371 228 11 refer refer VERB bracis-28371 228 12 to to ADP bracis-28371 228 13 lr lr NOUN bracis-28371 228 14 on on ADP bracis-28371 228 15 the the DET bracis-28371 228 16 \(\text \(\text NOUN bracis-28371 228 17 { { PUNCT bracis-28371 228 18 hmv}_{cti}\ hmv}_{cti}\ PROPN bracis-28371 228 19 ) ) PUNCT bracis-28371 228 20 dataset dataset NOUN bracis-28371 228 21 . . PUNCT bracis-28371 229 1 in in ADP bracis-28371 229 2 fig fig NOUN bracis-28371 229 3 . . PUNCT bracis-28371 229 4     SPACE bracis-28371 229 5 4f 4f NUM bracis-28371 229 6 , , PUNCT bracis-28371 229 7 we we PRON bracis-28371 229 8 can can AUX bracis-28371 229 9 observe observe VERB bracis-28371 229 10 that that SCONJ bracis-28371 229 11 patients patient NOUN bracis-28371 229 12 with with ADP bracis-28371 229 13 heart heart NOUN bracis-28371 229 14 disease disease NOUN bracis-28371 229 15 tend tend VERB bracis-28371 229 16 to to PART bracis-28371 229 17 be be AUX bracis-28371 229 18 older old ADJ bracis-28371 229 19 . . PUNCT bracis-28371 230 1 both both DET bracis-28371 230 2 features feature VERB bracis-28371 230 3 predispose predispose VERB bracis-28371 230 4 those those DET bracis-28371 230 5 patients patient NOUN bracis-28371 230 6 to to PART bracis-28371 230 7 be be AUX bracis-28371 230 8 classified classify VERB bracis-28371 230 9 in in ADP bracis-28371 230 10 the the DET bracis-28371 230 11 deceased deceased ADJ bracis-28371 230 12 class class NOUN bracis-28371 230 13 . . PUNCT bracis-28371 231 1 this this DET bracis-28371 231 2 finding finding NOUN bracis-28371 231 3 is be AUX bracis-28371 231 4 in in ADP bracis-28371 231 5 agreement agreement NOUN bracis-28371 231 6 with with ADP bracis-28371 231 7 the the DET bracis-28371 231 8 related relate VERB bracis-28371 231 9 literature literature NOUN bracis-28371 231 10 – – PUNCT bracis-28371 231 11 studies study NOUN bracis-28371 231 12 have have AUX bracis-28371 231 13 shown show VERB bracis-28371 231 14 that that SCONJ bracis-28371 231 15 heart heart NOUN bracis-28371 231 16 disease disease NOUN bracis-28371 231 17 is be AUX bracis-28371 231 18 a a DET bracis-28371 231 19 risk risk NOUN bracis-28371 231 20 factor factor NOUN bracis-28371 231 21 for for ADP bracis-28371 231 22 patient patient ADJ bracis-28371 231 23 mortality mortality NOUN bracis-28371 231 24     SPACE bracis-28371 232 1 [ [ X bracis-28371 232 2 4 4 NUM bracis-28371 232 3 ] ] PUNCT bracis-28371 232 4 . . PUNCT bracis-28371 233 1 comparing compare VERB bracis-28371 233 2 the the DET bracis-28371 233 3 predictions prediction NOUN bracis-28371 233 4 for for ADP bracis-28371 233 5 \(\text \(\text NOUN bracis-28371 233 6 { { PUNCT bracis-28371 233 7 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 233 8 ) ) PUNCT bracis-28371 233 9 using use VERB bracis-28371 233 10 xgb xgb PROPN bracis-28371 233 11 ( ( PUNCT bracis-28371 233 12 figs fig NOUN bracis-28371 233 13 . . PUNCT bracis-28371 233 14     SPACE bracis-28371 233 15 4 4 NUM bracis-28371 233 16 g g NOUN bracis-28371 233 17 , , PUNCT bracis-28371 233 18     SPACE bracis-28371 233 19 4h 4h NOUN bracis-28371 233 20 , , PUNCT bracis-28371 233 21 and and CCONJ bracis-28371 233 22     SPACE bracis-28371 233 23 4i 4i NOUN bracis-28371 233 24 ) ) PUNCT bracis-28371 233 25 and and CCONJ bracis-28371 233 26 lr lr INTJ bracis-28371 233 27 ( ( PUNCT bracis-28371 233 28 figs fig NOUN bracis-28371 233 29 . . PUNCT bracis-28371 233 30     SPACE bracis-28371 233 31 4j 4j NOUN bracis-28371 233 32 , , PUNCT bracis-28371 233 33     SPACE bracis-28371 233 34 4k 4k NOUN bracis-28371 233 35 , , PUNCT bracis-28371 233 36 and and CCONJ bracis-28371 233 37     SPACE bracis-28371 233 38 4l 4l NOUN bracis-28371 233 39 ) ) PUNCT bracis-28371 233 40 we we PRON bracis-28371 233 41 notice notice VERB bracis-28371 233 42 that that SCONJ bracis-28371 233 43 xgb xgb PROPN bracis-28371 233 44 tends tend VERB bracis-28371 233 45 to to PART bracis-28371 233 46 present present VERB bracis-28371 233 47 a a DET bracis-28371 233 48 greater great ADJ bracis-28371 233 49 overlap overlap NOUN bracis-28371 233 50 between between ADP bracis-28371 233 51 fn fn NOUN bracis-28371 233 52 and and CCONJ bracis-28371 233 53 fp fp NOUN bracis-28371 233 54 . . PROPN bracis-28371 234 1 this this PRON bracis-28371 234 2 may may AUX bracis-28371 234 3 be be AUX bracis-28371 234 4 due due ADJ bracis-28371 234 5 to to ADP bracis-28371 234 6 the the DET bracis-28371 234 7 influence influence NOUN bracis-28371 234 8 of of ADP bracis-28371 234 9 the the DET bracis-28371 234 10 t t PROPN bracis-28371 234 11 - - PUNCT bracis-28371 234 12 sne sne NOUN bracis-28371 234 13 visualization visualization NOUN bracis-28371 234 14 method method NOUN bracis-28371 234 15 , , PUNCT bracis-28371 234 16 which which PRON bracis-28371 234 17 may may AUX bracis-28371 234 18 favor favor VERB bracis-28371 234 19 the the DET bracis-28371 234 20 visualization visualization NOUN bracis-28371 234 21 for for ADP bracis-28371 234 22 the the DET bracis-28371 234 23 lr lr NOUN bracis-28371 234 24 algorithm algorithm NOUN bracis-28371 234 25 . . PUNCT bracis-28371 235 1 figures figure NOUN bracis-28371 235 2     SPACE bracis-28371 235 3 4i 4i NOUN bracis-28371 235 4 and and CCONJ bracis-28371 235 5     SPACE bracis-28371 235 6 4l 4l NOUN bracis-28371 235 7 , , PUNCT bracis-28371 235 8 generated generate VERB bracis-28371 235 9 based base VERB bracis-28371 235 10 on on ADP bracis-28371 235 11 the the DET bracis-28371 235 12 shap shap PROPN bracis-28371 235 13 values value NOUN bracis-28371 235 14 provided provide VERB bracis-28371 235 15 by by ADP bracis-28371 235 16 the the DET bracis-28371 235 17 xgb xgb PROPN bracis-28371 235 18 and and CCONJ bracis-28371 235 19 lr lr PROPN bracis-28371 235 20 classifiers classifier NOUN bracis-28371 235 21 , , PUNCT bracis-28371 235 22 show show VERB bracis-28371 235 23 regions region NOUN bracis-28371 235 24 of of ADP bracis-28371 235 25 high high ADJ bracis-28371 235 26 concentrations concentration NOUN bracis-28371 235 27 of of ADP bracis-28371 235 28 patients patient NOUN bracis-28371 235 29 with with ADP bracis-28371 235 30 high high ADJ bracis-28371 235 31 values value NOUN bracis-28371 235 32 of of ADP bracis-28371 235 33 d d NOUN bracis-28371 235 34 - - NOUN bracis-28371 235 35 dimers dimers NOUN bracis-28371 235 36 . . PUNCT bracis-28371 236 1 both both DET bracis-28371 236 2 younger young ADJ bracis-28371 236 3 and and CCONJ bracis-28371 236 4 older old ADJ bracis-28371 236 5 patients patient NOUN bracis-28371 236 6 are be AUX bracis-28371 236 7 found find VERB bracis-28371 236 8 in in ADP bracis-28371 236 9 those those DET bracis-28371 236 10 regions region NOUN bracis-28371 236 11 . . PUNCT bracis-28371 237 1 the the DET bracis-28371 237 2 pattern pattern NOUN bracis-28371 237 3 is be AUX bracis-28371 237 4 more more ADV bracis-28371 237 5 evident evident ADJ bracis-28371 237 6 in in ADP bracis-28371 237 7 the the DET bracis-28371 237 8 visualization visualization NOUN bracis-28371 237 9 obtained obtain VERB bracis-28371 237 10 with with ADP bracis-28371 237 11 lr lr NOUN bracis-28371 237 12 , , PUNCT bracis-28371 237 13 in in ADP bracis-28371 237 14 which which PRON bracis-28371 237 15 tp tp NOUN bracis-28371 237 16 and and CCONJ bracis-28371 237 17 fp fp PROPN bracis-28371 237 18 classifications classification NOUN bracis-28371 237 19 prevail prevail VERB bracis-28371 237 20 . . PUNCT bracis-28371 238 1 high high ADJ bracis-28371 238 2 d d ADJ bracis-28371 238 3 - - PUNCT bracis-28371 238 4 dimer dimer ADJ bracis-28371 238 5 values value NOUN bracis-28371 238 6 are be AUX bracis-28371 238 7 associated associate VERB bracis-28371 238 8 with with ADP bracis-28371 238 9 poor poor ADJ bracis-28371 238 10 prognosis prognosis NOUN bracis-28371 238 11 in in ADP bracis-28371 238 12 patients patient NOUN bracis-28371 238 13 with with ADP bracis-28371 238 14 covid-19 covid-19 PROPN bracis-28371 238 15 infection infection NOUN bracis-28371 238 16     SPACE bracis-28371 239 1 [ [ X bracis-28371 239 2 15 15 NUM bracis-28371 239 3 ] ] PUNCT bracis-28371 239 4 , , PUNCT bracis-28371 239 5 and and CCONJ bracis-28371 239 6 in in ADP bracis-28371 239 7 both both DET bracis-28371 239 8 cases case NOUN bracis-28371 239 9 ( ( PUNCT bracis-28371 239 10 xgb xgb NOUN bracis-28371 239 11 and and CCONJ bracis-28371 239 12 lr lr NOUN bracis-28371 239 13 ) ) PUNCT bracis-28371 239 14 , , PUNCT bracis-28371 239 15 models model NOUN bracis-28371 239 16 tend tend VERB bracis-28371 239 17 to to PART bracis-28371 239 18 assign assign VERB bracis-28371 239 19 these these DET bracis-28371 239 20 patients patient NOUN bracis-28371 239 21 to to ADP bracis-28371 239 22 the the DET bracis-28371 239 23 deceased deceased ADJ bracis-28371 239 24 class class NOUN bracis-28371 239 25 . . PUNCT bracis-28371 240 1 4.4 4.4 NUM bracis-28371 240 2 single single ADJ bracis-28371 240 3 vs. vs. ADP bracis-28371 240 4 multiple multiple ADJ bracis-28371 240 5 variable variable ADJ bracis-28371 240 6 classifiers classifier NOUN bracis-28371 240 7 ( ( PUNCT bracis-28371 240 8 rq3 rq3 PROPN bracis-28371 240 9 ) ) PUNCT bracis-28371 240 10 as as SCONJ bracis-28371 240 11 seen see VERB bracis-28371 240 12 in in ADP bracis-28371 240 13 the the DET bracis-28371 240 14 analysis analysis NOUN bracis-28371 240 15 of of ADP bracis-28371 240 16 sect sect NOUN bracis-28371 240 17 . . PUNCT bracis-28371 240 18     SPACE bracis-28371 240 19 4.3 4.3 NUM bracis-28371 240 20 , , PUNCT bracis-28371 240 21 as as ADV bracis-28371 240 22 well well ADV bracis-28371 240 23 as as ADP bracis-28371 240 24 recurrently recurrently ADV bracis-28371 240 25 pointed point VERB bracis-28371 240 26 out out ADP bracis-28371 240 27 by by ADP bracis-28371 240 28 works work NOUN bracis-28371 240 29 in in ADP bracis-28371 240 30 the the DET bracis-28371 240 31 related relate VERB bracis-28371 240 32 literature literature NOUN bracis-28371 240 33 , , PUNCT bracis-28371 240 34 age age NOUN bracis-28371 240 35 was be AUX bracis-28371 240 36 frequently frequently ADV bracis-28371 240 37 flagged flag VERB bracis-28371 240 38 as as ADP bracis-28371 240 39 the the DET bracis-28371 240 40 attribute attribute NOUN bracis-28371 240 41 with with ADP bracis-28371 240 42 the the DET bracis-28371 240 43 greatest great ADJ bracis-28371 240 44 contribution contribution NOUN bracis-28371 240 45 during during ADP bracis-28371 240 46 mortality mortality NOUN bracis-28371 240 47 prediction prediction NOUN bracis-28371 240 48 tasks task NOUN bracis-28371 240 49 . . PUNCT bracis-28371 241 1 this this DET bracis-28371 241 2 research research NOUN bracis-28371 241 3 question question NOUN bracis-28371 241 4 aims aim VERB bracis-28371 241 5 to to PART bracis-28371 241 6 compare compare VERB bracis-28371 241 7 the the DET bracis-28371 241 8 performance performance NOUN bracis-28371 241 9 between between ADP bracis-28371 241 10 classifiers classifier NOUN bracis-28371 241 11 with with ADP bracis-28371 241 12 multiple multiple ADJ bracis-28371 241 13 independent independent ADJ bracis-28371 241 14 variables variable NOUN bracis-28371 241 15 and and CCONJ bracis-28371 241 16 their their PRON bracis-28371 241 17 respective respective ADJ bracis-28371 241 18 versions version NOUN bracis-28371 241 19 with with ADP bracis-28371 241 20 only only ADV bracis-28371 241 21 one one NUM bracis-28371 241 22 independent independent ADJ bracis-28371 241 23 variable variable NOUN bracis-28371 241 24 . . PUNCT bracis-28371 242 1 for for ADP bracis-28371 242 2 this this PRON bracis-28371 242 3 , , PUNCT bracis-28371 242 4 we we PRON bracis-28371 242 5 compare compare VERB bracis-28371 242 6 the the DET bracis-28371 242 7 performance performance NOUN bracis-28371 242 8 of of ADP bracis-28371 242 9 the the DET bracis-28371 242 10 classifiers classifier NOUN bracis-28371 242 11 presented present VERB bracis-28371 242 12 in in ADP bracis-28371 242 13 sect sect NOUN bracis-28371 242 14 . . PUNCT bracis-28371 242 15     SPACE bracis-28371 243 1 4.2 4.2 NUM bracis-28371 243 2 . . PUNCT bracis-28371 244 1 figure figure VERB bracis-28371 244 2     SPACE bracis-28371 244 3 5 5 NUM bracis-28371 244 4 shows show VERB bracis-28371 244 5 the the DET bracis-28371 244 6 differences difference NOUN bracis-28371 244 7 between between ADP bracis-28371 244 8 the the DET bracis-28371 244 9 performances performance NOUN bracis-28371 244 10 obtained obtain VERB bracis-28371 244 11 by by ADP bracis-28371 244 12 the the DET bracis-28371 244 13 classifiers classifier NOUN bracis-28371 244 14 trained train VERB bracis-28371 244 15 using use VERB bracis-28371 244 16 only only ADJ bracis-28371 244 17 age age NOUN bracis-28371 244 18 as as ADP bracis-28371 244 19 independent independent ADJ bracis-28371 244 20 variable variable NOUN bracis-28371 244 21 . . PUNCT bracis-28371 245 1 fig fig NOUN bracis-28371 245 2 . . PUNCT bracis-28371 246 1 5 5 NUM bracis-28371 246 2 . . NOUN bracis-28371 246 3 multivariable multivariable ADJ bracis-28371 246 4 and and CCONJ bracis-28371 246 5 single single ADJ bracis-28371 246 6 - - PUNCT bracis-28371 246 7 variable variable ADJ bracis-28371 246 8 classifiers classifier NOUN bracis-28371 246 9 ( ( PUNCT bracis-28371 246 10 age age NOUN bracis-28371 246 11 ) ) PUNCT bracis-28371 246 12 for for ADP bracis-28371 246 13 mortality mortality NOUN bracis-28371 246 14 prediction prediction NOUN bracis-28371 246 15 full full ADJ bracis-28371 246 16 size size NOUN bracis-28371 246 17 image image NOUN bracis-28371 246 18 for for ADP bracis-28371 246 19 most most ADJ bracis-28371 246 20 algorithms algorithm NOUN bracis-28371 246 21 , , PUNCT bracis-28371 246 22 classification classification NOUN bracis-28371 246 23 models model NOUN bracis-28371 246 24 that that PRON bracis-28371 246 25 used use VERB bracis-28371 246 26 only only ADJ bracis-28371 246 27 age age NOUN bracis-28371 246 28 as as ADP bracis-28371 246 29 independent independent ADJ bracis-28371 246 30 variable variable NOUN bracis-28371 246 31 did do AUX bracis-28371 246 32 not not PART bracis-28371 246 33 show show VERB bracis-28371 246 34 degradation degradation NOUN bracis-28371 246 35 in in ADP bracis-28371 246 36 sensitivity sensitivity NOUN bracis-28371 246 37 , , PUNCT bracis-28371 246 38 and and CCONJ bracis-28371 246 39 in in ADP bracis-28371 246 40 some some DET bracis-28371 246 41 cases case NOUN bracis-28371 246 42 , , PUNCT bracis-28371 246 43 higher high ADJ bracis-28371 246 44 levels level NOUN bracis-28371 246 45 of of ADP bracis-28371 246 46 sensitivity sensitivity NOUN bracis-28371 246 47 were be AUX bracis-28371 246 48 found find VERB bracis-28371 246 49 . . PUNCT bracis-28371 247 1 although although SCONJ bracis-28371 247 2 this this DET bracis-28371 247 3 positive positive ADJ bracis-28371 247 4 effect effect NOUN bracis-28371 247 5 on on ADP bracis-28371 247 6 sensitivity sensitivity NOUN bracis-28371 247 7 levels level NOUN bracis-28371 247 8 was be AUX bracis-28371 247 9 observed observe VERB bracis-28371 247 10 , , PUNCT bracis-28371 247 11 the the DET bracis-28371 247 12 reduction reduction NOUN bracis-28371 247 13 in in ADP bracis-28371 247 14 ppv ppv NOUN bracis-28371 247 15 and and CCONJ bracis-28371 247 16 specificity specificity NOUN bracis-28371 247 17 levels level NOUN bracis-28371 247 18 was be AUX bracis-28371 247 19 also also ADV bracis-28371 247 20 recurrent recurrent ADJ bracis-28371 247 21 . . PUNCT bracis-28371 248 1 as as ADP bracis-28371 248 2 a a DET bracis-28371 248 3 consequence consequence NOUN bracis-28371 248 4 , , PUNCT bracis-28371 248 5 in in ADP bracis-28371 248 6 most most ADJ bracis-28371 248 7 cases case NOUN bracis-28371 248 8 , , PUNCT bracis-28371 248 9 no no DET bracis-28371 248 10 gains gain NOUN bracis-28371 248 11 were be AUX bracis-28371 248 12 observed observe VERB bracis-28371 248 13 for for ADP bracis-28371 248 14 the the DET bracis-28371 248 15 metrics metric NOUN bracis-28371 248 16 f1 f1 NOUN bracis-28371 248 17 + + PROPN bracis-28371 248 18 , , PUNCT bracis-28371 248 19 ma ma PROPN bracis-28371 248 20 - - PUNCT bracis-28371 248 21 f1 f1 NOUN bracis-28371 248 22 , , PUNCT bracis-28371 248 23 kappa kappa ADJ bracis-28371 248 24 , , PUNCT bracis-28371 248 25 auroc auroc NOUN bracis-28371 248 26 and and CCONJ bracis-28371 248 27 auprc auprc ADJ bracis-28371 248 28 . . PUNCT bracis-28371 249 1 the the DET bracis-28371 249 2 lr lr PROPN bracis-28371 249 3 algorithm algorithm NOUN bracis-28371 249 4 , , PUNCT bracis-28371 249 5 which which PRON bracis-28371 249 6 showed show VERB bracis-28371 249 7 the the DET bracis-28371 249 8 best good ADJ bracis-28371 249 9 results result NOUN bracis-28371 249 10 in in ADP bracis-28371 249 11 the the DET bracis-28371 249 12 experiments experiment NOUN bracis-28371 249 13 in in ADP bracis-28371 249 14 sect sect NOUN bracis-28371 249 15 . . PUNCT bracis-28371 249 16     SPACE bracis-28371 249 17 4.2 4.2 NUM bracis-28371 249 18 , , PUNCT bracis-28371 249 19 tended tend VERB bracis-28371 249 20 not not PART bracis-28371 249 21 to to PART bracis-28371 249 22 show show VERB bracis-28371 249 23 substantial substantial ADJ bracis-28371 249 24 decrease decrease NOUN bracis-28371 249 25 in in ADP bracis-28371 249 26 sensitivity sensitivity NOUN bracis-28371 249 27 levels level NOUN bracis-28371 249 28 . . PUNCT bracis-28371 250 1 for for ADP bracis-28371 250 2 the the DET bracis-28371 250 3 \(\text \(\text NOUN bracis-28371 250 4 { { PUNCT bracis-28371 250 5 hmv}_{all}\ hmv}_{all}\ PROPN bracis-28371 250 6 ) ) PUNCT bracis-28371 250 7 dataset dataset NOUN bracis-28371 250 8 , , PUNCT bracis-28371 250 9 considerable considerable ADJ bracis-28371 250 10 increases increase NOUN bracis-28371 250 11 in in ADP bracis-28371 250 12 sensitivity sensitivity NOUN bracis-28371 250 13 levels level NOUN bracis-28371 250 14 occur occur VERB bracis-28371 250 15 for for ADP bracis-28371 250 16 some some DET bracis-28371 250 17 algorithms algorithm NOUN bracis-28371 250 18 . . PUNCT bracis-28371 251 1 in in ADP bracis-28371 251 2 contrast contrast NOUN bracis-28371 251 3 , , PUNCT bracis-28371 251 4 there there PRON bracis-28371 251 5 were be VERB bracis-28371 251 6 considerable considerable ADJ bracis-28371 251 7 reductions reduction NOUN bracis-28371 251 8 in in ADP bracis-28371 251 9 the the DET bracis-28371 251 10 levels level NOUN bracis-28371 251 11 of of ADP bracis-28371 251 12 ppv ppv NOUN bracis-28371 251 13 and and CCONJ bracis-28371 251 14 specificity specificity NOUN bracis-28371 251 15 , , PUNCT bracis-28371 251 16 mainly mainly ADV bracis-28371 251 17 in in ADP bracis-28371 251 18 the the DET bracis-28371 251 19 \(\text \(\text NOUN bracis-28371 251 20 { { PUNCT bracis-28371 251 21 hmv}_{all}\ hmv}_{all}\ PROPN bracis-28371 251 22 ) ) PUNCT bracis-28371 251 23 dataset dataset NOUN bracis-28371 251 24 and and CCONJ bracis-28371 251 25 in in ADP bracis-28371 251 26 the the DET bracis-28371 251 27 \(\text \(\text NOUN bracis-28371 251 28 { { PUNCT bracis-28371 251 29 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 251 30 ) ) PUNCT bracis-28371 251 31 dataset dataset NOUN bracis-28371 251 32 . . PUNCT bracis-28371 252 1 algorithm algorithm PROPN bracis-28371 252 2 j48 j48 PROPN bracis-28371 252 3 , , PUNCT bracis-28371 252 4 which which PRON bracis-28371 252 5 presented present VERB bracis-28371 252 6 the the DET bracis-28371 252 7 lowest low ADJ bracis-28371 252 8 sensitivity sensitivity NOUN bracis-28371 252 9 scores score NOUN bracis-28371 252 10 in in ADP bracis-28371 252 11 the the DET bracis-28371 252 12 experiment experiment NOUN bracis-28371 252 13 in in ADP bracis-28371 252 14 sect sect NOUN bracis-28371 252 15 . . PUNCT bracis-28371 252 16     SPACE bracis-28371 252 17 4.2 4.2 NUM bracis-28371 252 18 , , PUNCT bracis-28371 252 19 was be AUX bracis-28371 252 20 recurrently recurrently ADV bracis-28371 252 21 the the DET bracis-28371 252 22 one one NOUN bracis-28371 252 23 that that PRON bracis-28371 252 24 obtained obtain VERB bracis-28371 252 25 the the DET bracis-28371 252 26 highest high ADJ bracis-28371 252 27 sensitivity sensitivity NOUN bracis-28371 252 28 gains gain NOUN bracis-28371 252 29 among among ADP bracis-28371 252 30 the the DET bracis-28371 252 31 classifiers classifier NOUN bracis-28371 252 32 that that PRON bracis-28371 252 33 used use VERB bracis-28371 252 34 only only ADJ bracis-28371 252 35 age age NOUN bracis-28371 252 36 as as ADP bracis-28371 252 37 an an DET bracis-28371 252 38 independent independent ADJ bracis-28371 252 39 variable variable NOUN bracis-28371 252 40 . . PUNCT bracis-28371 253 1 when when SCONJ bracis-28371 253 2 considering consider VERB bracis-28371 253 3 the the DET bracis-28371 253 4 \(\text \(\text ADJ bracis-28371 253 5 { { PUNCT bracis-28371 253 6 hmv}_{cti}\ hmv}_{cti}\ PROPN bracis-28371 253 7 ) ) PUNCT bracis-28371 253 8 dataset dataset NOUN bracis-28371 253 9 , , PUNCT bracis-28371 253 10 in in ADP bracis-28371 253 11 addition addition NOUN bracis-28371 253 12 to to ADP bracis-28371 253 13 the the DET bracis-28371 253 14 j48 j48 PROPN bracis-28371 253 15 algorithm algorithm NOUN bracis-28371 253 16 , , PUNCT bracis-28371 253 17 the the DET bracis-28371 253 18 rf rf NOUN bracis-28371 253 19 and and CCONJ bracis-28371 253 20 mlp mlp PROPN bracis-28371 253 21 algorithms algorithm NOUN bracis-28371 253 22 stood stand VERB bracis-28371 253 23 out out ADP bracis-28371 253 24 for for ADP bracis-28371 253 25 their their PRON bracis-28371 253 26 increased increase VERB bracis-28371 253 27 sensitivity sensitivity NOUN bracis-28371 253 28 . . PUNCT bracis-28371 254 1 however however ADV bracis-28371 254 2 , , PUNCT bracis-28371 254 3 despite despite SCONJ bracis-28371 254 4 the the DET bracis-28371 254 5 increase increase NOUN bracis-28371 254 6 in in ADP bracis-28371 254 7 sensitivity sensitivity NOUN bracis-28371 254 8 , , PUNCT bracis-28371 254 9 due due ADP bracis-28371 254 10 to to ADP bracis-28371 254 11 the the DET bracis-28371 254 12 losses loss NOUN bracis-28371 254 13 in in ADP bracis-28371 254 14 the the DET bracis-28371 254 15 levels level NOUN bracis-28371 254 16 of of ADP bracis-28371 254 17 ppv ppv NOUN bracis-28371 254 18 and and CCONJ bracis-28371 254 19 specificity specificity NOUN bracis-28371 254 20 , , PUNCT bracis-28371 254 21 no no DET bracis-28371 254 22 gains gain NOUN bracis-28371 254 23 were be AUX bracis-28371 254 24 observed observe VERB bracis-28371 254 25 for for ADP bracis-28371 254 26 the the DET bracis-28371 254 27 metrics metric NOUN bracis-28371 254 28 f1 f1 NOUN bracis-28371 254 29 + + PROPN bracis-28371 254 30 , , PUNCT bracis-28371 254 31 ma ma PROPN bracis-28371 254 32 - - PUNCT bracis-28371 254 33 f1 f1 PROPN bracis-28371 254 34 and and CCONJ bracis-28371 254 35 kappa kappa PROPN bracis-28371 254 36 . . PUNCT bracis-28371 255 1 finally finally ADV bracis-28371 255 2 , , PUNCT bracis-28371 255 3 j48 j48 PROPN bracis-28371 255 4 and and CCONJ bracis-28371 255 5 mlp mlp NOUN bracis-28371 255 6 were be AUX bracis-28371 255 7 the the DET bracis-28371 255 8 only only ADJ bracis-28371 255 9 ones one NOUN bracis-28371 255 10 that that PRON bracis-28371 255 11 obtained obtain VERB bracis-28371 255 12 auroc auroc NOUN bracis-28371 255 13 gains gain NOUN bracis-28371 255 14 . . PUNCT bracis-28371 256 1 in in ADP bracis-28371 256 2 \(\text \(\text PROPN bracis-28371 256 3 { { PUNCT bracis-28371 256 4 hcpa}_{cti}\ hcpa}_{cti}\ PROPN bracis-28371 256 5 ) ) PUNCT bracis-28371 256 6 ( ( PUNCT bracis-28371 256 7 fig fig NOUN bracis-28371 256 8 . . PUNCT bracis-28371 256 9     SPACE bracis-28371 256 10 5b 5b NUM bracis-28371 256 11 ) ) PUNCT bracis-28371 256 12 , , PUNCT bracis-28371 256 13 only only ADV bracis-28371 256 14 the the DET bracis-28371 256 15 j48 j48 PROPN bracis-28371 256 16 algorithm algorithm NOUN bracis-28371 256 17 obtained obtain VERB bracis-28371 256 18 an an DET bracis-28371 256 19 increase increase NOUN bracis-28371 256 20 in in ADP bracis-28371 256 21 the the DET bracis-28371 256 22 other other ADJ bracis-28371 256 23 metrics metric NOUN bracis-28371 256 24 , , PUNCT bracis-28371 256 25 in in ADP bracis-28371 256 26 addition addition NOUN bracis-28371 256 27 to to ADP bracis-28371 256 28 sensitivity sensitivity NOUN bracis-28371 256 29 . . PUNCT bracis-28371 257 1 in in ADP bracis-28371 257 2 \(\text \(\text PROPN bracis-28371 257 3 { { PUNCT bracis-28371 257 4 hmv}_{cti}\ hmv}_{cti}\ PROPN bracis-28371 257 5 ) ) PUNCT bracis-28371 257 6 ( ( PUNCT bracis-28371 257 7 fig fig NOUN bracis-28371 257 8 . . PUNCT bracis-28371 257 9     SPACE bracis-28371 257 10 5c 5c NUM bracis-28371 257 11 ) ) PUNCT bracis-28371 257 12 , , PUNCT bracis-28371 257 13 j48 j48 PROPN bracis-28371 257 14 and and CCONJ bracis-28371 257 15 mlp mlp PROPN bracis-28371 257 16 achieved achieve VERB bracis-28371 257 17 the the DET bracis-28371 257 18 highest high ADJ bracis-28371 257 19 sensitivity sensitivity NOUN bracis-28371 257 20 gains gain NOUN bracis-28371 257 21 . . PUNCT bracis-28371 258 1 in in ADP bracis-28371 258 2 this this DET bracis-28371 258 3 scenario scenario NOUN bracis-28371 258 4 , , PUNCT bracis-28371 258 5 both both DET bracis-28371 258 6 algorithms algorithm NOUN bracis-28371 258 7 showed show VERB bracis-28371 258 8 a a DET bracis-28371 258 9 smaller small ADJ bracis-28371 258 10 reduction reduction NOUN bracis-28371 258 11 in in ADP bracis-28371 258 12 terms term NOUN bracis-28371 258 13 of of ADP bracis-28371 258 14 ppv ppv NOUN bracis-28371 258 15 and and CCONJ bracis-28371 258 16 specificity specificity NOUN bracis-28371 258 17 , , PUNCT bracis-28371 258 18 resulting result VERB bracis-28371 258 19 in in ADP bracis-28371 258 20 an an DET bracis-28371 258 21 increase increase NOUN bracis-28371 258 22 in in ADP bracis-28371 258 23 the the DET bracis-28371 258 24 levels level NOUN bracis-28371 258 25 of of ADP bracis-28371 258 26 other other ADJ bracis-28371 258 27 metrics metric NOUN bracis-28371 258 28 . . PUNCT bracis-28371 259 1 essentially essentially ADV bracis-28371 259 2 , , PUNCT bracis-28371 259 3 the the DET bracis-28371 259 4 classifiers classifier NOUN bracis-28371 259 5 that that PRON bracis-28371 259 6 use use VERB bracis-28371 259 7 only only ADJ bracis-28371 259 8 age age NOUN bracis-28371 259 9 as as SCONJ bracis-28371 259 10 independent independent ADJ bracis-28371 259 11 variable variable NOUN bracis-28371 259 12 reach reach VERB bracis-28371 259 13 similar similar ADJ bracis-28371 259 14 or or CCONJ bracis-28371 259 15 higher high ADJ bracis-28371 259 16 sensitivity sensitivity NOUN bracis-28371 259 17 scores score NOUN bracis-28371 259 18 compared compare VERB bracis-28371 259 19 to to ADP bracis-28371 259 20 their their PRON bracis-28371 259 21 multivariable multivariable ADJ bracis-28371 259 22 versions version NOUN bracis-28371 259 23 . . PUNCT bracis-28371 260 1 however however ADV bracis-28371 260 2 , , PUNCT bracis-28371 260 3 these these DET bracis-28371 260 4 scores score NOUN bracis-28371 260 5 were be AUX bracis-28371 260 6 achieved achieve VERB bracis-28371 260 7 at at ADP bracis-28371 260 8 the the DET bracis-28371 260 9 cost cost NOUN bracis-28371 260 10 of of ADP bracis-28371 260 11 lower low ADJ bracis-28371 260 12 spe spe NOUN bracis-28371 260 13 and and CCONJ bracis-28371 260 14 ppv ppv NOUN bracis-28371 260 15 levels level NOUN bracis-28371 260 16 , , PUNCT bracis-28371 260 17 showing show VERB bracis-28371 260 18 that that SCONJ bracis-28371 260 19 other other ADJ bracis-28371 260 20 variables variable NOUN bracis-28371 260 21 are be AUX bracis-28371 260 22 useful useful ADJ bracis-28371 260 23 in in ADP bracis-28371 260 24 improving improve VERB bracis-28371 260 25 these these DET bracis-28371 260 26 metrics metric NOUN bracis-28371 260 27 . . PUNCT bracis-28371 261 1 the the DET bracis-28371 261 2 full full ADJ bracis-28371 261 3 version version NOUN bracis-28371 261 4 of of ADP bracis-28371 261 5 table table NOUN bracis-28371 261 6     SPACE bracis-28371 261 7 2 2 NUM bracis-28371 261 8 with with ADP bracis-28371 261 9 ci ci PROPN bracis-28371 261 10 ( ( PUNCT bracis-28371 261 11 rq1 rq1 NOUN bracis-28371 261 12 ) ) PUNCT bracis-28371 261 13 and and CCONJ bracis-28371 261 14 significance significance NOUN bracis-28371 261 15 tests test NOUN bracis-28371 261 16 performed perform VERB bracis-28371 261 17 on on ADP bracis-28371 261 18 rq3 rq3 NOUN bracis-28371 261 19 can can AUX bracis-28371 261 20 be be AUX bracis-28371 261 21 accessed access VERB bracis-28371 261 22 at at ADP bracis-28371 261 23 https://github.com/danimtk/covid-19-paper https://github.com/danimtk/covid-19-paper NOUN bracis-28371 261 24 . . PUNCT bracis-28371 262 1 5 5 NUM bracis-28371 262 2 conclusion conclusion NOUN bracis-28371 262 3 this this DET bracis-28371 262 4 work work NOUN bracis-28371 262 5 evaluated evaluate VERB bracis-28371 262 6 several several ADJ bracis-28371 262 7 ml ml NOUN bracis-28371 262 8 algorithms algorithm NOUN bracis-28371 262 9 for for ADP bracis-28371 262 10 predicting predict VERB bracis-28371 262 11 covid-19 covid-19 PROPN bracis-28371 262 12 mortality mortality NOUN bracis-28371 262 13 . . PUNCT bracis-28371 263 1 the the DET bracis-28371 263 2 classifiers classifier NOUN bracis-28371 263 3 showed show VERB bracis-28371 263 4 a a DET bracis-28371 263 5 promising promising ADJ bracis-28371 263 6 ability ability NOUN bracis-28371 263 7 to to PART bracis-28371 263 8 predict predict VERB bracis-28371 263 9 mortality mortality NOUN bracis-28371 263 10 based base VERB bracis-28371 263 11 on on ADP bracis-28371 263 12 the the DET bracis-28371 263 13 information information NOUN bracis-28371 263 14 available available ADJ bracis-28371 263 15 at at ADP bracis-28371 263 16 the the DET bracis-28371 263 17 time time NOUN bracis-28371 263 18 of of ADP bracis-28371 263 19 admission admission NOUN bracis-28371 263 20 , , PUNCT bracis-28371 263 21 reaching reach VERB bracis-28371 263 22 a a DET bracis-28371 263 23 sensitivity sensitivity NOUN bracis-28371 263 24 of of ADP bracis-28371 263 25 up up ADP bracis-28371 263 26 to to PART bracis-28371 263 27 84 84 NUM bracis-28371 263 28 % % NOUN bracis-28371 263 29 and and CCONJ bracis-28371 263 30 an an DET bracis-28371 263 31 auroc auroc NOUN bracis-28371 263 32 of of ADP bracis-28371 263 33 92 92 NUM bracis-28371 263 34 % % NOUN bracis-28371 263 35 ( ( PUNCT bracis-28371 263 36 95 95 NUM bracis-28371 263 37 % % NOUN bracis-28371 263 38 ci ci NOUN bracis-28371 263 39 , , PUNCT bracis-28371 263 40 [ [ X bracis-28371 263 41 0.89–0.95 0.89–0.95 NOUN bracis-28371 263 42 ] ] NUM bracis-28371 263 43 ) ) PUNCT bracis-28371 263 44 . . PUNCT bracis-28371 264 1 overall overall ADJ bracis-28371 264 2 , , PUNCT bracis-28371 264 3 besides besides SCONJ bracis-28371 264 4 age age NOUN bracis-28371 264 5 , , PUNCT bracis-28371 264 6 oxygen oxygen NOUN bracis-28371 264 7 saturation saturation NOUN bracis-28371 264 8 , , PUNCT bracis-28371 264 9 heart heart NOUN bracis-28371 264 10 disease disease NOUN bracis-28371 264 11 , , PUNCT bracis-28371 264 12 hemogram hemogram PROPN bracis-28371 264 13 information information NOUN bracis-28371 264 14 also also ADV bracis-28371 264 15 tended tend VERB bracis-28371 264 16 to to PART bracis-28371 264 17 have have VERB bracis-28371 264 18 high high ADJ bracis-28371 264 19 contributions contribution NOUN bracis-28371 264 20 to to ADP bracis-28371 264 21 the the DET bracis-28371 264 22 predictions prediction NOUN bracis-28371 264 23 . . PUNCT bracis-28371 265 1 we we PRON bracis-28371 265 2 applied apply VERB bracis-28371 265 3 a a DET bracis-28371 265 4 visualization visualization NOUN bracis-28371 265 5 technique technique NOUN bracis-28371 265 6 based base VERB bracis-28371 265 7 on on ADP bracis-28371 265 8 shap shap PROPN bracis-28371 265 9 values value NOUN bracis-28371 265 10 , , PUNCT bracis-28371 265 11 which which PRON bracis-28371 265 12 helps help VERB bracis-28371 265 13 to to PART bracis-28371 265 14 corroborate corroborate VERB bracis-28371 265 15 the the DET bracis-28371 265 16 high high ADJ bracis-28371 265 17 importance importance NOUN bracis-28371 265 18 attributed attribute VERB bracis-28371 265 19 to to ADP bracis-28371 265 20 the the DET bracis-28371 265 21 age age NOUN bracis-28371 265 22 feature feature NOUN bracis-28371 265 23 by by ADP bracis-28371 265 24 the the DET bracis-28371 265 25 classifiers classifier NOUN bracis-28371 265 26 in in ADP bracis-28371 265 27 their their PRON bracis-28371 265 28 predictions prediction NOUN bracis-28371 265 29 . . PUNCT bracis-28371 266 1 the the DET bracis-28371 266 2 visualizations visualization NOUN bracis-28371 266 3 also also ADV bracis-28371 266 4 showed show VERB bracis-28371 266 5 that that SCONJ bracis-28371 266 6 some some DET bracis-28371 266 7 groups group NOUN bracis-28371 266 8 of of ADP bracis-28371 266 9 patients patient NOUN bracis-28371 266 10 with with ADP bracis-28371 266 11 low low ADJ bracis-28371 266 12 erythrocytes erythrocyte NOUN bracis-28371 266 13 and and CCONJ bracis-28371 266 14 high high ADJ bracis-28371 266 15 d d NOUN bracis-28371 266 16 - - PUNCT bracis-28371 266 17 dimer dimer NOUN bracis-28371 266 18 ( ( PUNCT bracis-28371 266 19 risk risk NOUN bracis-28371 266 20 factors factor NOUN bracis-28371 266 21 for for ADP bracis-28371 266 22 mortality mortality NOUN bracis-28371 266 23 ) ) PUNCT bracis-28371 266 24 tend tend VERB bracis-28371 266 25 to to PART bracis-28371 266 26 be be AUX bracis-28371 266 27 recognized recognize VERB bracis-28371 266 28 by by ADP bracis-28371 266 29 the the DET bracis-28371 266 30 classifiers classifier NOUN bracis-28371 266 31 with with ADP bracis-28371 266 32 a a DET bracis-28371 266 33 higher high ADJ bracis-28371 266 34 risk risk NOUN bracis-28371 266 35 of of ADP bracis-28371 266 36 dying die VERB bracis-28371 266 37 . . PUNCT bracis-28371 267 1 we we PRON bracis-28371 267 2 believe believe VERB bracis-28371 267 3 that that SCONJ bracis-28371 267 4 the the DET bracis-28371 267 5 visualization visualization NOUN bracis-28371 267 6 technique technique NOUN bracis-28371 267 7 employed employ VERB bracis-28371 267 8 here here ADV bracis-28371 267 9 can can AUX bracis-28371 267 10 also also ADV bracis-28371 267 11 be be AUX bracis-28371 267 12 used use VERB bracis-28371 267 13 in in ADP bracis-28371 267 14 other other ADJ bracis-28371 267 15 prediction prediction NOUN bracis-28371 267 16 tasks task NOUN bracis-28371 267 17 . . PUNCT bracis-28371 268 1 one one NUM bracis-28371 268 2 limitation limitation NOUN bracis-28371 268 3 of of ADP bracis-28371 268 4 our our PRON bracis-28371 268 5 work work NOUN bracis-28371 268 6 is be AUX bracis-28371 268 7 that that SCONJ bracis-28371 268 8 shap shap PROPN bracis-28371 268 9 is be AUX bracis-28371 268 10 relatively relatively ADV bracis-28371 268 11 time time NOUN bracis-28371 268 12 - - PUNCT bracis-28371 268 13 consuming consume VERB bracis-28371 268 14 when when SCONJ bracis-28371 268 15 applied apply VERB bracis-28371 268 16 to to ADP bracis-28371 268 17 datasets dataset NOUN bracis-28371 268 18 with with ADP bracis-28371 268 19 a a DET bracis-28371 268 20 large large ADJ bracis-28371 268 21 number number NOUN bracis-28371 268 22 of of ADP bracis-28371 268 23 features feature NOUN bracis-28371 268 24     SPACE bracis-28371 269 1 [ [ X bracis-28371 269 2 19 19 NUM bracis-28371 269 3 ] ] PUNCT bracis-28371 269 4 . . PUNCT bracis-28371 270 1 this this DET bracis-28371 270 2 limitation limitation NOUN bracis-28371 270 3 can can AUX bracis-28371 270 4 be be AUX bracis-28371 270 5 mitigated mitigate VERB bracis-28371 270 6 using use VERB bracis-28371 270 7 optimized optimize VERB bracis-28371 270 8 shap shap NOUN bracis-28371 270 9 algorithms algorithm NOUN bracis-28371 270 10 ( ( PUNCT bracis-28371 270 11 like like INTJ bracis-28371 270 12 treeshap treeshap NOUN bracis-28371 270 13 or or CCONJ bracis-28371 270 14 [ [ X bracis-28371 270 15 19 19 NUM bracis-28371 270 16 ] ] NUM bracis-28371 270 17 ) ) PUNCT bracis-28371 270 18 . . PUNCT bracis-28371 271 1 although although SCONJ bracis-28371 271 2 the the DET bracis-28371 271 3 visualizations visualization NOUN bracis-28371 271 4 obtained obtain VERB bracis-28371 271 5 are be AUX bracis-28371 271 6 intuitive intuitive ADJ bracis-28371 271 7 and and CCONJ bracis-28371 271 8 seem seem VERB bracis-28371 271 9 to to PART bracis-28371 271 10 represent represent VERB bracis-28371 271 11 already already ADV bracis-28371 271 12 known know VERB bracis-28371 271 13 relationships relationship NOUN bracis-28371 271 14 ( ( PUNCT bracis-28371 271 15 such such ADJ bracis-28371 271 16 as as ADP bracis-28371 271 17 the the DET bracis-28371 271 18 high high ADJ bracis-28371 271 19 risk risk NOUN bracis-28371 271 20 for for ADP bracis-28371 271 21 older old ADJ bracis-28371 271 22 patients patient NOUN bracis-28371 271 23 , , PUNCT bracis-28371 271 24 as as ADV bracis-28371 271 25 well well ADV bracis-28371 271 26 as as ADP bracis-28371 271 27 those those PRON bracis-28371 271 28 with with ADP bracis-28371 271 29 comorbidities comorbiditie NOUN bracis-28371 271 30 ) ) PUNCT bracis-28371 271 31 , , PUNCT bracis-28371 271 32 it it PRON bracis-28371 271 33 should should AUX bracis-28371 271 34 be be AUX bracis-28371 271 35 noted note VERB bracis-28371 271 36 that that SCONJ bracis-28371 271 37 approaches approach NOUN bracis-28371 271 38 that that PRON bracis-28371 271 39 use use VERB bracis-28371 271 40 shapley shapley ADJ bracis-28371 271 41 values value NOUN bracis-28371 271 42 , , PUNCT bracis-28371 271 43 as as ADV bracis-28371 271 44 well well ADV bracis-28371 271 45 as as ADP bracis-28371 271 46 approximations approximation NOUN bracis-28371 271 47 of of ADP bracis-28371 271 48 these these DET bracis-28371 271 49 values value NOUN bracis-28371 271 50 ( ( PUNCT bracis-28371 271 51 such such ADJ bracis-28371 271 52 as as ADP bracis-28371 271 53 shap shap NOUN bracis-28371 271 54 ) ) PUNCT bracis-28371 271 55 , , PUNCT bracis-28371 271 56 may may AUX bracis-28371 271 57 generate generate VERB bracis-28371 271 58 misconceptions misconception NOUN bracis-28371 271 59 regarding regard VERB bracis-28371 271 60 the the DET bracis-28371 271 61 actual actual ADJ bracis-28371 271 62 contribution contribution NOUN bracis-28371 271 63 of of ADP bracis-28371 271 64 features feature NOUN bracis-28371 271 65 [ [ X bracis-28371 271 66 8 8 NUM bracis-28371 271 67 ] ] PUNCT bracis-28371 271 68 . . PUNCT bracis-28371 272 1 thus thus ADV bracis-28371 272 2 , , PUNCT bracis-28371 272 3 we we PRON bracis-28371 272 4 emphasize emphasize VERB bracis-28371 272 5 the the DET bracis-28371 272 6 importance importance NOUN bracis-28371 272 7 for for SCONJ bracis-28371 272 8 specialists specialist NOUN bracis-28371 272 9 to to PART bracis-28371 272 10 use use VERB bracis-28371 272 11 other other ADJ bracis-28371 272 12 evaluation evaluation NOUN bracis-28371 272 13 techniques technique NOUN bracis-28371 272 14 concomitantly concomitantly ADV bracis-28371 272 15 to to PART bracis-28371 272 16 assist assist VERB bracis-28371 272 17 in in ADP bracis-28371 272 18 the the DET bracis-28371 272 19 interpretation interpretation NOUN bracis-28371 272 20 of of ADP bracis-28371 272 21 predictive predictive ADJ bracis-28371 272 22 models model NOUN bracis-28371 272 23 , , PUNCT bracis-28371 272 24 avoiding avoid VERB bracis-28371 272 25 misinterpretations misinterpretation NOUN bracis-28371 272 26 . . PUNCT bracis-28371 273 1 in in ADP bracis-28371 273 2 future future ADJ bracis-28371 273 3 work work NOUN bracis-28371 273 4 , , PUNCT bracis-28371 273 5 we we PRON bracis-28371 273 6 plan plan VERB bracis-28371 273 7 to to PART bracis-28371 273 8 experiment experiment VERB bracis-28371 273 9 with with ADP bracis-28371 273 10 other other ADJ bracis-28371 273 11 dimensionality dimensionality NOUN bracis-28371 273 12 reduction reduction NOUN bracis-28371 273 13 methods method NOUN bracis-28371 273 14 and and CCONJ bracis-28371 273 15 classifiers classifier NOUN bracis-28371 273 16 . . PUNCT bracis-28371 274 1 we we PRON bracis-28371 274 2 also also ADV bracis-28371 274 3 plan plan VERB bracis-28371 274 4 to to PART bracis-28371 274 5 investigate investigate VERB bracis-28371 274 6 whether whether SCONJ bracis-28371 274 7 there there PRON bracis-28371 274 8 is be VERB bracis-28371 274 9 a a DET bracis-28371 274 10 set set NOUN bracis-28371 274 11 of of ADP bracis-28371 274 12 attributes attribute NOUN bracis-28371 274 13 ( ( PUNCT bracis-28371 274 14 and and CCONJ bracis-28371 274 15 how how SCONJ bracis-28371 274 16 many many ADJ bracis-28371 274 17 ) ) PUNCT bracis-28371 274 18 would would AUX bracis-28371 274 19 be be AUX bracis-28371 274 20 needed need VERB bracis-28371 274 21 to to PART bracis-28371 274 22 maintain maintain VERB bracis-28371 274 23 sensitivity sensitivity NOUN bracis-28371 274 24 levels level NOUN bracis-28371 274 25 without without ADP bracis-28371 274 26 considerable considerable ADJ bracis-28371 274 27 ppv ppv NOUN bracis-28371 274 28 losses loss NOUN bracis-28371 274 29 . . PUNCT bracis-28371 275 1 notes note VERB bracis-28371 275 2 1.ethics 1.ethics NUM bracis-28371 275 3 committee committee NOUN bracis-28371 275 4 approval approval NOUN bracis-28371 275 5 : : PUNCT bracis-28371 275 6 hcpa hcpa VERB bracis-28371 275 7 32314720.8.0000.5327 32314720.8.0000.5327 NUM bracis-28371 275 8 , , PUNCT bracis-28371 275 9 hmv hmv PROPN bracis-28371 275 10 32314720.8.3001.5330 32314720.8.3001.5330 PROPN bracis-28371 275 11 . . PUNCT bracis-28371 276 1 2.https://fracpete.github.io/python-weka-wrapper3 2.https://fracpete.github.io/python-weka-wrapper3 NOUN bracis-28371 276 2 . . PUNCT bracis-28371 277 1 3.https://github.com/dmlc/xgboost 3.https://github.com/dmlc/xgboost NUM bracis-28371 277 2 . . PUNCT bracis-28371 278 1 4.https://github.com/slundberg/shap 4.https://github.com/slundberg/shap NUM bracis-28371 278 2 . . PUNCT bracis-28371 279 1 references reference NOUN bracis-28371 279 2 aas aas PROPN bracis-28371 279 3 , , PUNCT bracis-28371 279 4 k. k. PROPN bracis-28371 279 5 , , PUNCT bracis-28371 279 6 jullum jullum PROPN bracis-28371 279 7 , , PUNCT bracis-28371 279 8 m. m. NOUN bracis-28371 279 9 , , PUNCT bracis-28371 279 10 løland løland NOUN bracis-28371 279 11 , , PUNCT bracis-28371 279 12 a. a. NOUN bracis-28371 279 13 : : PUNCT bracis-28371 279 14 explaining explain VERB bracis-28371 279 15 individual individual ADJ bracis-28371 279 16 predictions prediction NOUN bracis-28371 279 17 when when SCONJ bracis-28371 279 18 features feature NOUN bracis-28371 279 19 are be AUX bracis-28371 279 20 dependent dependent ADJ bracis-28371 279 21 : : PUNCT bracis-28371 279 22 accurate accurate ADJ bracis-28371 279 23 approximations approximation NOUN bracis-28371 279 24 to to ADP bracis-28371 279 25 shapley shapley NOUN bracis-28371 279 26 values value NOUN bracis-28371 279 27 . . PUNCT bracis-28371 280 1 artif artif INTJ bracis-28371 280 2 . . PUNCT bracis-28371 281 1 intell intell PROPN bracis-28371 281 2 . . PUNCT bracis-28371 282 1 298 298 NUM bracis-28371 282 2 , , PUNCT bracis-28371 282 3 103502 103502 NUM bracis-28371 282 4 ( ( PUNCT bracis-28371 282 5 2021 2021 NUM bracis-28371 282 6 ) ) PUNCT bracis-28371 282 7 article article NOUN bracis-28371 282 8     SPACE bracis-28371 282 9 mathscinet mathscinet NOUN bracis-28371 282 10     SPACE bracis-28371 282 11 math math NOUN bracis-28371 282 12     SPACE bracis-28371 282 13 google google PROPN bracis-28371 282 14 scholar scholar NOUN bracis-28371 282 15     SPACE bracis-28371 282 16 alballa alballa NOUN bracis-28371 282 17 , , PUNCT bracis-28371 282 18 n. n. PROPN bracis-28371 282 19 , , PUNCT bracis-28371 282 20 al al PROPN bracis-28371 282 21 - - PUNCT bracis-28371 282 22 turaiki turaiki PROPN bracis-28371 282 23 , , PUNCT bracis-28371 282 24 i. i. NOUN bracis-28371 282 25 : : PUNCT bracis-28371 282 26 machine machine NOUN bracis-28371 282 27 learning learn VERB bracis-28371 282 28 approaches approach NOUN bracis-28371 282 29 in in ADP bracis-28371 282 30 covid-19 covid-19 PROPN bracis-28371 282 31 diagnosis diagnosis NOUN bracis-28371 282 32 , , PUNCT bracis-28371 282 33 mortality mortality NOUN bracis-28371 282 34 , , PUNCT bracis-28371 282 35 and and CCONJ bracis-28371 282 36 severity severity NOUN bracis-28371 282 37 risk risk NOUN bracis-28371 282 38 prediction prediction NOUN bracis-28371 282 39 : : PUNCT bracis-28371 282 40 a a DET bracis-28371 282 41 review review NOUN bracis-28371 282 42 . . PUNCT bracis-28371 283 1 inf inf PROPN bracis-28371 283 2 . . PUNCT bracis-28371 283 3 med med PROPN bracis-28371 283 4 . . PUNCT bracis-28371 284 1 unlocked unlock VERB bracis-28371 284 2 24 24 NUM bracis-28371 284 3 , , PUNCT bracis-28371 284 4 100564 100564 NUM bracis-28371 284 5 ( ( PUNCT bracis-28371 284 6 2021 2021 NUM bracis-28371 284 7 ) ) PUNCT bracis-28371 284 8 article article NOUN bracis-28371 284 9     SPACE bracis-28371 284 10 google google PROPN bracis-28371 284 11 scholar scholar NOUN bracis-28371 284 12     SPACE bracis-28371 284 13 araújo araújo PROPN bracis-28371 284 14 , , PUNCT bracis-28371 284 15 d.c d.c PROPN bracis-28371 284 16 . . PROPN bracis-28371 284 17 , , PUNCT bracis-28371 284 18 veloso veloso PROPN bracis-28371 284 19 , , PUNCT bracis-28371 284 20 a.a a.a PROPN bracis-28371 284 21 . . PROPN bracis-28371 284 22 , , PUNCT bracis-28371 284 23 borges borges PROPN bracis-28371 284 24 , , PUNCT bracis-28371 284 25 k.b.g k.b.g PROPN bracis-28371 284 26 . . PROPN bracis-28371 284 27 , , PUNCT bracis-28371 284 28 das das PROPN bracis-28371 284 29 graças graças PROPN bracis-28371 284 30 carvalho carvalho PROPN bracis-28371 284 31 , , PUNCT bracis-28371 284 32 m. m. NOUN bracis-28371 284 33 : : PUNCT bracis-28371 284 34 prognosing prognose VERB bracis-28371 284 35 the the DET bracis-28371 284 36 risk risk NOUN bracis-28371 284 37 of of ADP bracis-28371 284 38 covid-19 covid-19 PROPN bracis-28371 284 39 death death NOUN bracis-28371 284 40 through through ADP bracis-28371 284 41 a a DET bracis-28371 284 42 machine machine NOUN bracis-28371 284 43 learning learning NOUN bracis-28371 284 44 - - PUNCT bracis-28371 284 45 based base VERB bracis-28371 284 46 routine routine ADJ bracis-28371 284 47 blood blood NOUN bracis-28371 284 48 panel panel NOUN bracis-28371 284 49 : : PUNCT bracis-28371 284 50 a a DET bracis-28371 284 51 retrospective retrospective ADJ bracis-28371 284 52 study study NOUN bracis-28371 284 53 in in ADP bracis-28371 284 54 brazil brazil PROPN bracis-28371 284 55 . . PUNCT bracis-28371 285 1 ijmedi ijmedi PROPN bracis-28371 285 2 165 165 NUM bracis-28371 285 3 , , PUNCT bracis-28371 285 4 104835 104835 NUM bracis-28371 285 5 ( ( PUNCT bracis-28371 285 6 2022 2022 NUM bracis-28371 285 7 ) ) PUNCT bracis-28371 285 8 google google PROPN bracis-28371 285 9 scholar scholar NOUN bracis-28371 285 10     SPACE bracis-28371 285 11 broberg broberg PROPN bracis-28371 285 12 , , PUNCT bracis-28371 285 13 c.s c.s PROPN bracis-28371 285 14 . . PROPN bracis-28371 285 15 , , PUNCT bracis-28371 285 16 kovacs kovacs PROPN bracis-28371 285 17 , , PUNCT bracis-28371 285 18 a.h a.h PROPN bracis-28371 285 19 . . PROPN bracis-28371 285 20 , , PUNCT bracis-28371 285 21 sadeghi sadeghi PROPN bracis-28371 285 22 , , PUNCT bracis-28371 285 23 s. s. PROPN bracis-28371 285 24 , , PUNCT bracis-28371 285 25 et et PROPN bracis-28371 285 26 al al PROPN bracis-28371 285 27 . . PUNCT bracis-28371 285 28 : : PUNCT bracis-28371 286 1 covid-19 covid-19 PROPN bracis-28371 286 2 in in ADP bracis-28371 286 3 adults adult NOUN bracis-28371 286 4 with with ADP bracis-28371 286 5 congenital congenital ADJ bracis-28371 286 6 heart heart NOUN bracis-28371 286 7 disease disease NOUN bracis-28371 286 8 . . PUNCT bracis-28371 287 1 jacc jacc PROPN bracis-28371 287 2 77(13 77(13 PROPN bracis-28371 287 3 ) ) PUNCT bracis-28371 287 4 , , PUNCT bracis-28371 287 5 1644–1655 1644–1655 NUM bracis-28371 287 6 ( ( PUNCT bracis-28371 287 7 2021 2021 NUM bracis-28371 287 8 ) ) PUNCT bracis-28371 287 9 article article NOUN bracis-28371 287 10     SPACE bracis-28371 287 11 google google PROPN bracis-28371 287 12 scholar scholar NOUN bracis-28371 287 13     SPACE bracis-28371 287 14 covino covino PROPN bracis-28371 287 15 , , PUNCT bracis-28371 287 16 m. m. NOUN bracis-28371 287 17 , , PUNCT bracis-28371 287 18 sandroni sandroni ADJ bracis-28371 287 19 , , PUNCT bracis-28371 287 20 c. c. PROPN bracis-28371 287 21 , , PUNCT bracis-28371 287 22 santoro santoro PROPN bracis-28371 287 23 , , PUNCT bracis-28371 287 24 m. m. NOUN bracis-28371 287 25 , , PUNCT bracis-28371 287 26 et et PROPN bracis-28371 287 27 al al PROPN bracis-28371 287 28 . . PROPN bracis-28371 287 29 : : PUNCT bracis-28371 288 1 predicting predict VERB bracis-28371 288 2 intensive intensive ADJ bracis-28371 288 3 care care NOUN bracis-28371 288 4 unit unit NOUN bracis-28371 288 5 admission admission NOUN bracis-28371 288 6 and and CCONJ bracis-28371 288 7 death death NOUN bracis-28371 288 8 for for ADP bracis-28371 288 9 covid-19 covid-19 PROPN bracis-28371 288 10 patients patient NOUN bracis-28371 288 11 in in ADP bracis-28371 288 12 the the DET bracis-28371 288 13 emergency emergency NOUN bracis-28371 288 14 department department NOUN bracis-28371 288 15 using use VERB bracis-28371 288 16 early early ADJ bracis-28371 288 17 warning warning NOUN bracis-28371 288 18 scores score NOUN bracis-28371 288 19 . . PUNCT bracis-28371 289 1 resuscitation resuscitation NOUN bracis-28371 289 2 156 156 NUM bracis-28371 289 3 , , PUNCT bracis-28371 289 4 84–91 84–91 NUM bracis-28371 289 5 ( ( PUNCT bracis-28371 289 6 2020 2020 NUM bracis-28371 289 7 ) ) PUNCT bracis-28371 289 8 article article NOUN bracis-28371 289 9     SPACE bracis-28371 289 10 google google PROPN bracis-28371 289 11 scholar scholar NOUN bracis-28371 289 12     SPACE bracis-28371 289 13 futoma futoma NOUN bracis-28371 289 14 , , PUNCT bracis-28371 289 15 j. j. PROPN bracis-28371 289 16 , , PUNCT bracis-28371 289 17 simons simon NOUN bracis-28371 289 18 , , PUNCT bracis-28371 289 19 m. m. NOUN bracis-28371 289 20 , , PUNCT bracis-28371 289 21 panch panch NOUN bracis-28371 289 22 , , PUNCT bracis-28371 289 23 t. t. PROPN bracis-28371 289 24 , , PUNCT bracis-28371 289 25 et et PROPN bracis-28371 289 26 al al PROPN bracis-28371 289 27 . . PUNCT bracis-28371 289 28 : : PUNCT bracis-28371 290 1 the the DET bracis-28371 290 2 myth myth NOUN bracis-28371 290 3 of of ADP bracis-28371 290 4 generalisability generalisability NOUN bracis-28371 290 5 in in ADP bracis-28371 290 6 clinical clinical ADJ bracis-28371 290 7 research research NOUN bracis-28371 290 8 and and CCONJ bracis-28371 290 9 machine machine NOUN bracis-28371 290 10 learning learning NOUN bracis-28371 290 11 in in ADP bracis-28371 290 12 health health NOUN bracis-28371 290 13 care care NOUN bracis-28371 290 14 . . PUNCT bracis-28371 291 1 lancet lancet NOUN bracis-28371 291 2 digit digit PROPN bracis-28371 291 3 . . PUNCT bracis-28371 292 1 health health NOUN bracis-28371 292 2 2(9 2(9 NUM bracis-28371 292 3 ) ) PUNCT bracis-28371 292 4 , , PUNCT bracis-28371 292 5 489–492 489–492 NUM bracis-28371 292 6 ( ( PUNCT bracis-28371 292 7 2020 2020 NUM bracis-28371 292 8 ) ) PUNCT bracis-28371 292 9 article article NOUN bracis-28371 292 10     SPACE bracis-28371 292 11 google google PROPN bracis-28371 292 12 scholar scholar NOUN bracis-28371 292 13     SPACE bracis-28371 292 14 hall hall NOUN bracis-28371 292 15 , , PUNCT bracis-28371 292 16 m.a m.a PROPN bracis-28371 292 17 . . PROPN bracis-28371 292 18 : : PUNCT bracis-28371 292 19 correlation correlation NOUN bracis-28371 292 20 - - PUNCT bracis-28371 292 21 based base VERB bracis-28371 292 22 feature feature NOUN bracis-28371 292 23 selection selection NOUN bracis-28371 292 24 for for ADP bracis-28371 292 25 machine machine NOUN bracis-28371 292 26 learning learning NOUN bracis-28371 292 27 . . PUNCT bracis-28371 293 1 ph.d ph.d PROPN bracis-28371 293 2 . . PUNCT bracis-28371 294 1 thesis thesis NOUN bracis-28371 294 2 , , PUNCT bracis-28371 294 3 the the DET bracis-28371 294 4 university university NOUN bracis-28371 294 5 of of ADP bracis-28371 294 6 waikato waikato PROPN bracis-28371 294 7 ( ( PUNCT bracis-28371 294 8 1999 1999 NUM bracis-28371 294 9 ) ) PUNCT bracis-28371 294 10 google google PROPN bracis-28371 294 11 scholar scholar NOUN bracis-28371 294 12     SPACE bracis-28371 294 13 huang huang PROPN bracis-28371 294 14 , , PUNCT bracis-28371 294 15 x. x. PROPN bracis-28371 294 16 , , PUNCT bracis-28371 294 17 marques marques PROPN bracis-28371 294 18 - - PROPN bracis-28371 294 19 silva silva PROPN bracis-28371 294 20 , , PUNCT bracis-28371 294 21 j. j. PROPN bracis-28371 294 22 : : PUNCT bracis-28371 294 23 the the DET bracis-28371 294 24 inadequacy inadequacy NOUN bracis-28371 294 25 of of ADP bracis-28371 294 26 shapley shapley NOUN bracis-28371 294 27 values value NOUN bracis-28371 294 28 for for ADP bracis-28371 294 29 explainability explainability NOUN bracis-28371 294 30 . . PUNCT bracis-28371 295 1 arxiv arxiv PROPN bracis-28371 295 2 preprint preprint PROPN bracis-28371 295 3 arxiv:2302.08160 arxiv:2302.08160 PROPN bracis-28371 295 4 ( ( PUNCT bracis-28371 295 5 2023 2023 NUM bracis-28371 295 6 ) ) PUNCT bracis-28371 295 7 knight knight NOUN bracis-28371 295 8 , , PUNCT bracis-28371 295 9 s.r s.r PROPN bracis-28371 295 10 . . PROPN bracis-28371 295 11 , , PUNCT bracis-28371 295 12 ho ho PROPN bracis-28371 295 13 , , PUNCT bracis-28371 295 14 a. a. PROPN bracis-28371 295 15 , , PUNCT bracis-28371 295 16 pius pius PROPN bracis-28371 295 17 , , PUNCT bracis-28371 295 18 r. r. PROPN bracis-28371 295 19 , , PUNCT bracis-28371 295 20 et et PROPN bracis-28371 295 21 al al PROPN bracis-28371 295 22 . . PUNCT bracis-28371 295 23 : : PUNCT bracis-28371 296 1 risk risk NOUN bracis-28371 296 2 stratification stratification NOUN bracis-28371 296 3 of of ADP bracis-28371 296 4 patients patient NOUN bracis-28371 296 5 admitted admit VERB bracis-28371 296 6 to to ADP bracis-28371 296 7 hospital hospital NOUN bracis-28371 296 8 with with ADP bracis-28371 296 9 covid-19 covid-19 PROPN bracis-28371 296 10 using use VERB bracis-28371 296 11 the the DET bracis-28371 296 12 isaric isaric NOUN bracis-28371 296 13 who who PRON bracis-28371 296 14 clinical clinical ADJ bracis-28371 296 15 characterisation characterisation NOUN bracis-28371 296 16 protocol protocol NOUN bracis-28371 296 17 : : PUNCT bracis-28371 296 18 development development NOUN bracis-28371 296 19 and and CCONJ bracis-28371 296 20 validation validation NOUN bracis-28371 296 21 of of ADP bracis-28371 296 22 the the DET bracis-28371 296 23 4c 4c NUM bracis-28371 296 24 mortality mortality NOUN bracis-28371 296 25 score score NOUN bracis-28371 296 26 . . PUNCT bracis-28371 297 1 bmj bmj PROPN bracis-28371 297 2 370 370 NUM bracis-28371 297 3 , , PUNCT bracis-28371 297 4 m3339 m3339 PROPN bracis-28371 297 5 ( ( PUNCT bracis-28371 297 6 2020 2020 NUM bracis-28371 297 7 ) ) PUNCT bracis-28371 297 8 article article NOUN bracis-28371 297 9     SPACE bracis-28371 297 10 google google PROPN bracis-28371 297 11 scholar scholar NOUN bracis-28371 297 12     SPACE bracis-28371 297 13 lundberg lundberg PROPN bracis-28371 297 14 , , PUNCT bracis-28371 297 15 s.m s.m PROPN bracis-28371 297 16 . . PROPN bracis-28371 297 17 , , PUNCT bracis-28371 297 18 erion erion NOUN bracis-28371 297 19 , , PUNCT bracis-28371 297 20 g.g g.g PROPN bracis-28371 297 21 . . PROPN bracis-28371 297 22 , , PUNCT bracis-28371 297 23 lee lee PROPN bracis-28371 297 24 , , PUNCT bracis-28371 297 25 s.i s.i PROPN bracis-28371 297 26 . . PROPN bracis-28371 297 27 : : PUNCT bracis-28371 298 1 consistent consistent ADJ bracis-28371 298 2 individualized individualized ADJ bracis-28371 298 3 feature feature NOUN bracis-28371 298 4 attribution attribution NOUN bracis-28371 298 5 for for ADP bracis-28371 298 6 tree tree NOUN bracis-28371 298 7 ensembles ensemble NOUN bracis-28371 298 8 . . PUNCT bracis-28371 299 1 arxiv arxiv PROPN bracis-28371 299 2 preprint preprint VERB bracis-28371 299 3 arxiv:1802.03888 arxiv:1802.03888 NOUN bracis-28371 299 4 ( ( PUNCT bracis-28371 299 5 2018 2018 NUM bracis-28371 299 6 ) ) PUNCT bracis-28371 299 7 lundberg lundberg PROPN bracis-28371 299 8 , , PUNCT bracis-28371 299 9 s.m s.m PROPN bracis-28371 299 10 . . PROPN bracis-28371 299 11 , , PUNCT bracis-28371 299 12 lee lee PROPN bracis-28371 299 13 , , PUNCT bracis-28371 299 14 s.i s.i PROPN bracis-28371 299 15 . . PROPN bracis-28371 299 16 : : PUNCT bracis-28371 300 1 a a DET bracis-28371 300 2 unified unified ADJ bracis-28371 300 3 approach approach NOUN bracis-28371 300 4 to to ADP bracis-28371 300 5 interpreting interpret VERB bracis-28371 300 6 model model NOUN bracis-28371 300 7 predictions prediction NOUN bracis-28371 300 8 . . PUNCT bracis-28371 301 1 in in ADP bracis-28371 301 2 : : PUNCT bracis-28371 301 3 advances advance NOUN bracis-28371 301 4 in in ADP bracis-28371 301 5 neural neural ADJ bracis-28371 301 6 information information NOUN bracis-28371 301 7 processing processing NOUN bracis-28371 301 8 systems system NOUN bracis-28371 301 9 , , PUNCT bracis-28371 301 10 vol vol NOUN bracis-28371 301 11 . . PROPN bracis-28371 301 12 30 30 NUM bracis-28371 301 13 ( ( PUNCT bracis-28371 301 14 2017 2017 NUM bracis-28371 301 15 ) ) PUNCT bracis-28371 301 16 google google PROPN bracis-28371 301 17 scholar scholar NOUN bracis-28371 301 18     SPACE bracis-28371 301 19 van van PROPN bracis-28371 301 20 der der NOUN bracis-28371 301 21 maaten maaten VERB bracis-28371 301 22 , , PUNCT bracis-28371 301 23 l. l. PROPN bracis-28371 301 24 , , PUNCT bracis-28371 301 25 hinton hinton PROPN bracis-28371 301 26 , , PUNCT bracis-28371 301 27 g. g. PROPN bracis-28371 301 28 : : PUNCT bracis-28371 301 29 visualizing visualize VERB bracis-28371 301 30 data datum NOUN bracis-28371 301 31 using use VERB bracis-28371 301 32 t t PROPN bracis-28371 301 33 - - PUNCT bracis-28371 301 34 sne sne NOUN bracis-28371 301 35 . . PUNCT bracis-28371 302 1 jmlr jmlr PROPN bracis-28371 302 2 9(11 9(11 NUM bracis-28371 302 3 ) ) PUNCT bracis-28371 302 4 , , PUNCT bracis-28371 302 5 2579–2605 2579–2605 NUM bracis-28371 302 6 ( ( PUNCT bracis-28371 302 7 2008 2008 NUM bracis-28371 302 8 ) ) PUNCT bracis-28371 302 9 math math NOUN bracis-28371 302 10     SPACE bracis-28371 302 11 google google PROPN bracis-28371 302 12 scholar scholar NOUN bracis-28371 302 13     SPACE bracis-28371 302 14 miranda miranda PROPN bracis-28371 302 15 de de PROPN bracis-28371 302 16 paiva paiva PROPN bracis-28371 302 17 , , PUNCT bracis-28371 302 18 b.b b.b PROPN bracis-28371 302 19 . . PROPN bracis-28371 302 20 , , PUNCT bracis-28371 302 21 delfino delfino PROPN bracis-28371 302 22 - - PUNCT bracis-28371 302 23 pereira pereira PROPN bracis-28371 302 24 , , PUNCT bracis-28371 302 25 p. p. PROPN bracis-28371 302 26 , , PUNCT bracis-28371 302 27 de de PROPN bracis-28371 302 28 andrade andrade PROPN bracis-28371 302 29 , , PUNCT bracis-28371 302 30 c.m.v c.m.v NOUN bracis-28371 302 31 . . PROPN bracis-28371 302 32 , , PUNCT bracis-28371 302 33 et et PROPN bracis-28371 302 34 al al PROPN bracis-28371 302 35 . . PROPN bracis-28371 302 36 : : PUNCT bracis-28371 303 1 effectiveness effectiveness NOUN bracis-28371 303 2 , , PUNCT bracis-28371 303 3 explainability explainability NOUN bracis-28371 303 4 and and CCONJ bracis-28371 303 5 reliability reliability NOUN bracis-28371 303 6 of of ADP bracis-28371 303 7 machine machine NOUN bracis-28371 303 8 meta meta ADV bracis-28371 303 9 - - PUNCT bracis-28371 303 10 learning learn VERB bracis-28371 303 11 methods method NOUN bracis-28371 303 12 for for ADP bracis-28371 303 13 predicting predict VERB bracis-28371 303 14 mortality mortality NOUN bracis-28371 303 15 in in ADP bracis-28371 303 16 patients patient NOUN bracis-28371 303 17 with with ADP bracis-28371 303 18 covid-19 covid-19 PROPN bracis-28371 303 19 : : PUNCT bracis-28371 303 20 results result NOUN bracis-28371 303 21 of of ADP bracis-28371 303 22 the the DET bracis-28371 303 23 brazilian brazilian ADJ bracis-28371 303 24 covid-19 covid-19 PROPN bracis-28371 303 25 registry registry NOUN bracis-28371 303 26 . . PUNCT bracis-28371 304 1 medrxiv medrxiv PROPN bracis-28371 304 2 ( ( PUNCT bracis-28371 304 3 2021 2021 NUM bracis-28371 304 4 ) ) PUNCT bracis-28371 304 5 google google NOUN bracis-28371 304 6 scholar scholar NOUN bracis-28371 304 7     SPACE bracis-28371 304 8 qin qin PROPN bracis-28371 304 9 , , PUNCT bracis-28371 304 10 z. z. PROPN bracis-28371 304 11 , , PUNCT bracis-28371 304 12 zhang zhang PROPN bracis-28371 304 13 , , PUNCT bracis-28371 304 14 c. c. PROPN bracis-28371 304 15 , , PUNCT bracis-28371 304 16 wang wang PROPN bracis-28371 304 17 , , PUNCT bracis-28371 304 18 t. t. PROPN bracis-28371 304 19 , , PUNCT bracis-28371 304 20 zhang zhang PROPN bracis-28371 304 21 , , PUNCT bracis-28371 304 22 s. s. PROPN bracis-28371 304 23 : : PUNCT bracis-28371 304 24 cost cost VERB bracis-28371 304 25 sensitive sensitive ADJ bracis-28371 304 26 classification classification NOUN bracis-28371 304 27 in in ADP bracis-28371 304 28 data datum NOUN bracis-28371 304 29 mining mining NOUN bracis-28371 304 30 . . PUNCT bracis-28371 305 1 in in ADP bracis-28371 305 2 : : PUNCT bracis-28371 305 3 advanced advanced ADJ bracis-28371 305 4 data data NOUN bracis-28371 305 5 mining mining NOUN bracis-28371 305 6 and and CCONJ bracis-28371 305 7 applications application NOUN bracis-28371 305 8 , , PUNCT bracis-28371 305 9 pp pp ADP bracis-28371 305 10 . . PUNCT bracis-28371 306 1 1–11 1–11 PROPN bracis-28371 306 2 ( ( PUNCT bracis-28371 306 3 2010 2010 NUM bracis-28371 306 4 ) ) PUNCT bracis-28371 306 5 google google PROPN bracis-28371 306 6 scholar scholar NOUN bracis-28371 306 7     SPACE bracis-28371 306 8 rostami rostami NOUN bracis-28371 306 9 , , PUNCT bracis-28371 306 10 m. m. NOUN bracis-28371 306 11 , , PUNCT bracis-28371 306 12 mansouritorghabeh mansouritorghabeh NOUN bracis-28371 306 13 , , PUNCT bracis-28371 306 14 h. h. NOUN bracis-28371 306 15 : : PUNCT bracis-28371 306 16 d d X bracis-28371 306 17 - - PUNCT bracis-28371 306 18 dimer dimer ADJ bracis-28371 306 19 level level NOUN bracis-28371 306 20 in in ADP bracis-28371 306 21 covid-19 covid-19 PROPN bracis-28371 306 22 infection infection NOUN bracis-28371 306 23 : : PUNCT bracis-28371 306 24 a a DET bracis-28371 306 25 systematic systematic ADJ bracis-28371 306 26 review review NOUN bracis-28371 306 27 . . PUNCT bracis-28371 307 1 exp exp PROPN bracis-28371 307 2 . . PROPN bracis-28371 307 3 rev rev PROPN bracis-28371 307 4 . . PROPN bracis-28371 307 5 hematol hematol NOUN bracis-28371 307 6 . . PUNCT bracis-28371 308 1 13(11 13(11 NUM bracis-28371 308 2 ) ) PUNCT bracis-28371 308 3 , , PUNCT bracis-28371 308 4 1265–1275 1265–1275 NUM bracis-28371 308 5 ( ( PUNCT bracis-28371 308 6 2020 2020 NUM bracis-28371 308 7 ) ) PUNCT bracis-28371 308 8 article article NOUN bracis-28371 308 9     SPACE bracis-28371 308 10 google google PROPN bracis-28371 308 11 scholar scholar NOUN bracis-28371 308 12     SPACE bracis-28371 308 13 subudhi subudhi PROPN bracis-28371 308 14 , , PUNCT bracis-28371 308 15 s. s. PROPN bracis-28371 308 16 , , PUNCT bracis-28371 308 17 verma verma PROPN bracis-28371 308 18 , , PUNCT bracis-28371 308 19 a. a. PROPN bracis-28371 308 20 , , PUNCT bracis-28371 308 21 patel patel PROPN bracis-28371 308 22 , , PUNCT bracis-28371 308 23 a.b a.b PROPN bracis-28371 308 24 . . PROPN bracis-28371 308 25 : : PUNCT bracis-28371 309 1 prognostic prognostic ADJ bracis-28371 309 2 machine machine NOUN bracis-28371 309 3 learning learning NOUN bracis-28371 309 4 models model NOUN bracis-28371 309 5 for for ADP bracis-28371 309 6 covid-19 covid-19 PROPN bracis-28371 309 7 to to PART bracis-28371 309 8 facilitate facilitate VERB bracis-28371 309 9 decision decision NOUN bracis-28371 309 10 making making NOUN bracis-28371 309 11 . . PUNCT bracis-28371 310 1 ijcp ijcp VERB bracis-28371 310 2 74(12 74(12 NUM bracis-28371 310 3 ) ) PUNCT bracis-28371 310 4 , , PUNCT bracis-28371 310 5 e13685 e13685 PROPN bracis-28371 310 6 ( ( PUNCT bracis-28371 310 7 2020 2020 NUM bracis-28371 310 8 ) ) PUNCT bracis-28371 310 9 google google PROPN bracis-28371 310 10 scholar scholar NOUN bracis-28371 310 11     SPACE bracis-28371 310 12 wynants wynant NOUN bracis-28371 310 13 , , PUNCT bracis-28371 310 14 l. l. PROPN bracis-28371 310 15 , , PUNCT bracis-28371 310 16 van van PROPN bracis-28371 310 17 calster calster PROPN bracis-28371 310 18 , , PUNCT bracis-28371 310 19 b. b. PROPN bracis-28371 310 20 , , PUNCT bracis-28371 310 21 collins collins PROPN bracis-28371 310 22 , , PUNCT bracis-28371 310 23 g.s g.s PROPN bracis-28371 310 24 . . PROPN bracis-28371 310 25 , , PUNCT bracis-28371 310 26 et et PROPN bracis-28371 310 27 al al PROPN bracis-28371 310 28 . . PROPN bracis-28371 310 29 : : PUNCT bracis-28371 310 30 prediction prediction NOUN bracis-28371 310 31 models model NOUN bracis-28371 310 32 for for ADP bracis-28371 310 33 diagnosis diagnosis NOUN bracis-28371 310 34 and and CCONJ bracis-28371 310 35 prognosis prognosis NOUN bracis-28371 310 36 of of ADP bracis-28371 310 37 covid-19 covid-19 PROPN bracis-28371 310 38 : : PUNCT bracis-28371 310 39 systematic systematic ADJ bracis-28371 310 40 review review NOUN bracis-28371 310 41 and and CCONJ bracis-28371 310 42 critical critical ADJ bracis-28371 310 43 appraisal appraisal NOUN bracis-28371 310 44 . . PUNCT bracis-28371 311 1 bmj bmj PROPN bracis-28371 311 2 369 369 NUM bracis-28371 311 3 , , PUNCT bracis-28371 311 4 m1328 m1328 PROPN bracis-28371 311 5 ( ( PUNCT bracis-28371 311 6 2020 2020 NUM bracis-28371 311 7 ) ) PUNCT bracis-28371 311 8 article article NOUN bracis-28371 311 9     SPACE bracis-28371 311 10 google google PROPN bracis-28371 311 11 scholar scholar NOUN bracis-28371 311 12     SPACE bracis-28371 311 13 yadaw yadaw NOUN bracis-28371 311 14 , , PUNCT bracis-28371 311 15 a.s a.s PROPN bracis-28371 311 16 . . PROPN bracis-28371 311 17 , , PUNCT bracis-28371 311 18 li li PROPN bracis-28371 311 19 , , PUNCT bracis-28371 311 20 y. y. PROPN bracis-28371 311 21 , , PUNCT bracis-28371 311 22 bose bose PROPN bracis-28371 311 23 , , PUNCT bracis-28371 311 24 s. s. PROPN bracis-28371 311 25 , , PUNCT bracis-28371 311 26 et et PROPN bracis-28371 311 27 al al PROPN bracis-28371 311 28 . . PUNCT bracis-28371 311 29 : : PUNCT bracis-28371 312 1 clinical clinical ADJ bracis-28371 312 2 features feature NOUN bracis-28371 312 3 of of ADP bracis-28371 312 4 covid-19 covid-19 PROPN bracis-28371 312 5 mortality mortality NOUN bracis-28371 312 6 : : PUNCT bracis-28371 312 7 development development NOUN bracis-28371 312 8 and and CCONJ bracis-28371 312 9 validation validation NOUN bracis-28371 312 10 of of ADP bracis-28371 312 11 a a DET bracis-28371 312 12 clinical clinical ADJ bracis-28371 312 13 prediction prediction NOUN bracis-28371 312 14 model model NOUN bracis-28371 312 15 . . PUNCT bracis-28371 313 1 lancet lancet PROPN bracis-28371 313 2 digt digt PROPN bracis-28371 313 3 . . PUNCT bracis-28371 314 1 health health NOUN bracis-28371 314 2 2(10 2(10 NUM bracis-28371 314 3 ) ) PUNCT bracis-28371 314 4 , , PUNCT bracis-28371 314 5 e516 e516 NUM bracis-28371 314 6 – – PUNCT bracis-28371 314 7 e525 e525 PROPN bracis-28371 314 8 ( ( PUNCT bracis-28371 314 9 2020 2020 NUM bracis-28371 314 10 ) ) PUNCT bracis-28371 314 11 article article NOUN bracis-28371 314 12     SPACE bracis-28371 314 13 google google PROPN bracis-28371 314 14 scholar scholar NOUN bracis-28371 314 15     SPACE bracis-28371 314 16 yang yang PROPN bracis-28371 314 17 , , PUNCT bracis-28371 314 18 j. j. PROPN bracis-28371 314 19 : : PUNCT bracis-28371 314 20 fast fast ADJ bracis-28371 314 21 treeshap treeshap ADV bracis-28371 314 22 : : PUNCT bracis-28371 314 23 accelerating accelerate VERB bracis-28371 314 24 shap shap PROPN bracis-28371 314 25 value value NOUN bracis-28371 314 26 computation computation NOUN bracis-28371 314 27 for for ADP bracis-28371 314 28 trees tree NOUN bracis-28371 314 29 . . PUNCT bracis-28371 315 1 arxiv arxiv PROPN bracis-28371 315 2 preprint preprint NOUN bracis-28371 315 3 arxiv:2109.09847 arxiv:2109.09847 NOUN bracis-28371 315 4 ( ( PUNCT bracis-28371 315 5 2021 2021 NUM bracis-28371 315 6 ) ) PUNCT bracis-28371 315 7 zhao zhao PROPN bracis-28371 315 8 , , PUNCT bracis-28371 315 9 z. z. PROPN bracis-28371 315 10 , , PUNCT bracis-28371 315 11 chen chen PROPN bracis-28371 315 12 , , PUNCT bracis-28371 315 13 a. a. PROPN bracis-28371 315 14 , , PUNCT bracis-28371 315 15 hou hou PROPN bracis-28371 315 16 , , PUNCT bracis-28371 315 17 w. w. PROPN bracis-28371 315 18 , , PUNCT bracis-28371 315 19 et et PROPN bracis-28371 315 20 al al PROPN bracis-28371 315 21 . . PROPN bracis-28371 315 22 : : PUNCT bracis-28371 316 1 prediction prediction NOUN bracis-28371 316 2 model model NOUN bracis-28371 316 3 and and CCONJ bracis-28371 316 4 risk risk NOUN bracis-28371 316 5 scores score NOUN bracis-28371 316 6 of of ADP bracis-28371 316 7 icu icu PROPN bracis-28371 316 8 admission admission NOUN bracis-28371 316 9 and and CCONJ bracis-28371 316 10 mortality mortality NOUN bracis-28371 316 11 in in ADP bracis-28371 316 12 covid-19 covid-19 PROPN bracis-28371 316 13 . . PUNCT bracis-28371 317 1 plos plos PROPN bracis-28371 317 2 one one NUM bracis-28371 317 3 15(7 15(7 NUM bracis-28371 317 4 ) ) PUNCT bracis-28371 317 5 , , PUNCT bracis-28371 317 6 e0236618 e0236618 PROPN bracis-28371 317 7 ( ( PUNCT bracis-28371 317 8 2020 2020 NUM bracis-28371 317 9 ) ) PUNCT bracis-28371 317 10 article article NOUN bracis-28371 317 11     SPACE bracis-28371 317 12 google google PROPN bracis-28371 317 13 scholar scholar NOUN bracis-28371 317 14     SPACE bracis-28371 317 15 download download NOUN bracis-28371 317 16 references reference NOUN bracis-28371 317 17 acknowledgment acknowledgment NOUN bracis-28371 317 18 this this DET bracis-28371 317 19 work work NOUN bracis-28371 317 20 has have AUX bracis-28371 317 21 been be AUX bracis-28371 317 22 financed finance VERB bracis-28371 317 23 in in ADP bracis-28371 317 24 part part NOUN bracis-28371 317 25 by by ADP bracis-28371 317 26 capes cape NOUN bracis-28371 317 27 finance finance NOUN bracis-28371 317 28 code code NOUN bracis-28371 317 29 001 001 NUM bracis-28371 317 30 and and CCONJ bracis-28371 317 31 cnpq cnpq NOUN bracis-28371 317 32 / / SYM bracis-28371 317 33 brazil brazil PROPN bracis-28371 317 34 . . PUNCT bracis-28371 318 1 author author NOUN bracis-28371 318 2 information information NOUN bracis-28371 318 3 authors author NOUN bracis-28371 318 4 and and CCONJ bracis-28371 318 5 affiliations affiliation NOUN bracis-28371 318 6 institute institute PROPN bracis-28371 318 7 of of ADP bracis-28371 318 8 informatics informatics PROPN bracis-28371 318 9 , , PUNCT bracis-28371 318 10 ufrgs ufrgs PROPN bracis-28371 318 11 , , PUNCT bracis-28371 318 12 porto porto PROPN bracis-28371 318 13 alegre alegre PROPN bracis-28371 318 14 , , PUNCT bracis-28371 318 15 brazil brazil PROPN bracis-28371 318 16 daniel daniel PROPN bracis-28371 318 17 matheus matheus PROPN bracis-28371 318 18 kuhn kuhn PROPN bracis-28371 318 19 , , PUNCT bracis-28371 318 20     SPACE bracis-28371 318 21 mariana mariana PROPN bracis-28371 318 22 recamonde recamonde PROPN bracis-28371 318 23 - - PUNCT bracis-28371 318 24 mendoza mendoza PROPN bracis-28371 318 25 , , PUNCT bracis-28371 318 26     SPACE bracis-28371 318 27 joão joão PROPN bracis-28371 318 28 luiz luiz PROPN bracis-28371 318 29 dihl dihl PROPN bracis-28371 318 30 comba comba PROPN bracis-28371 318 31     SPACE bracis-28371 318 32 & & CCONJ bracis-28371 318 33     SPACE bracis-28371 318 34 viviane viviane PROPN bracis-28371 318 35 pereira pereira PROPN bracis-28371 318 36 moreira moreira PROPN bracis-28371 318 37 hospital hospital PROPN bracis-28371 318 38 de de X bracis-28371 318 39 clínicas clínicas X bracis-28371 318 40 de de PROPN bracis-28371 318 41 porto porto PROPN bracis-28371 318 42 alegre alegre PROPN bracis-28371 318 43 , , PUNCT bracis-28371 318 44 porto porto PROPN bracis-28371 318 45 alegre alegre PROPN bracis-28371 318 46 , , PUNCT bracis-28371 318 47 brazil brazil PROPN bracis-28371 318 48 melina melina PROPN bracis-28371 318 49 silva silva PROPN bracis-28371 318 50 de de PROPN bracis-28371 318 51 loreto loreto PROPN bracis-28371 318 52 authors author NOUN bracis-28371 318 53 daniel daniel PROPN bracis-28371 318 54 matheus matheus PROPN bracis-28371 318 55 kuhnview kuhnview PROPN bracis-28371 318 56 author author NOUN bracis-28371 318 57 publications publication NOUN bracis-28371 318 58 search search NOUN bracis-28371 318 59 author author NOUN bracis-28371 318 60 on on ADP bracis-28371 318 61 : : PUNCT bracis-28371 318 62 pubmed pubmed PROPN bracis-28371 318 63     SPACE bracis-28371 318 64 google google PROPN bracis-28371 318 65 scholar scholar PROPN bracis-28371 318 66 melina melina PROPN bracis-28371 318 67 silva silva PROPN bracis-28371 318 68 de de PROPN bracis-28371 318 69 loretoview loretoview PROPN bracis-28371 318 70 author author NOUN bracis-28371 318 71 publications publication NOUN bracis-28371 318 72 search search NOUN bracis-28371 318 73 author author NOUN bracis-28371 318 74 on on ADP bracis-28371 318 75 : : PUNCT bracis-28371 318 76 pubmed pubmed PROPN bracis-28371 318 77     SPACE bracis-28371 318 78 google google PROPN bracis-28371 318 79 scholar scholar NOUN bracis-28371 318 80 mariana mariana PROPN bracis-28371 318 81 recamonde recamonde PROPN bracis-28371 318 82 - - PUNCT bracis-28371 318 83 mendozaview mendozaview NOUN bracis-28371 318 84 author author NOUN bracis-28371 318 85 publications publication NOUN bracis-28371 318 86 search search NOUN bracis-28371 318 87 author author NOUN bracis-28371 318 88 on on ADP bracis-28371 318 89 : : PUNCT bracis-28371 318 90 pubmed pubmed PROPN bracis-28371 318 91     SPACE bracis-28371 318 92 google google PROPN bracis-28371 318 93 scholar scholar PROPN bracis-28371 318 94 joão joão PROPN bracis-28371 318 95 luiz luiz PROPN bracis-28371 318 96 dihl dihl PROPN bracis-28371 318 97 combaview combaview PROPN bracis-28371 318 98 author author NOUN bracis-28371 318 99 publications publication NOUN bracis-28371 318 100 search search NOUN bracis-28371 318 101 author author NOUN bracis-28371 318 102 on on ADP bracis-28371 318 103 : : PUNCT bracis-28371 318 104 pubmed pubmed PROPN bracis-28371 318 105     SPACE bracis-28371 318 106 google google PROPN bracis-28371 318 107 scholar scholar NOUN bracis-28371 318 108 viviane viviane PROPN bracis-28371 318 109 pereira pereira PROPN bracis-28371 318 110 moreiraview moreiraview PROPN bracis-28371 318 111 author author NOUN bracis-28371 318 112 publications publication NOUN bracis-28371 318 113 search search NOUN bracis-28371 318 114 author author NOUN bracis-28371 318 115 on on ADP bracis-28371 318 116 : : PUNCT bracis-28371 318 117 pubmed pubmed PROPN bracis-28371 318 118     SPACE bracis-28371 318 119 google google PROPN bracis-28371 318 120 scholar scholar NOUN bracis-28371 318 121 corresponding correspond VERB bracis-28371 318 122 author author NOUN bracis-28371 318 123 correspondence correspondence NOUN bracis-28371 318 124 to to ADP bracis-28371 318 125 daniel daniel PROPN bracis-28371 318 126 matheus matheus PROPN bracis-28371 318 127 kuhn kuhn PROPN bracis-28371 318 128 . . PUNCT bracis-28371 319 1 editor editor NOUN bracis-28371 319 2 information information NOUN bracis-28371 319 3 editors editor NOUN bracis-28371 319 4 and and CCONJ bracis-28371 319 5 affiliations affiliation NOUN bracis-28371 319 6 federal federal PROPN bracis-28371 319 7 university university PROPN bracis-28371 319 8 of of ADP bracis-28371 319 9 são são PROPN bracis-28371 319 10 carlos carlos PROPN bracis-28371 319 11 , , PUNCT bracis-28371 319 12 são são PROPN bracis-28371 319 13 carlos carlos PROPN bracis-28371 319 14 , , PUNCT bracis-28371 319 15 brazil brazil PROPN bracis-28371 319 16 murilo murilo PROPN bracis-28371 319 17 c. c. PROPN bracis-28371 319 18 naldi naldi PROPN bracis-28371 319 19 centro centro PROPN bracis-28371 319 20 universitario universitario PROPN bracis-28371 319 21 da da PROPN bracis-28371 319 22 fei fei PROPN bracis-28371 319 23 , , PUNCT bracis-28371 319 24 são são PROPN bracis-28371 319 25 bernardo bernardo PROPN bracis-28371 319 26 do do AUX bracis-28371 319 27 campo campo PROPN bracis-28371 319 28 , , PUNCT bracis-28371 319 29 brazil brazil PROPN bracis-28371 319 30 reinaldo reinaldo PROPN bracis-28371 319 31 a. a. PROPN bracis-28371 319 32 c. c. PROPN bracis-28371 319 33 bianchi bianchi PROPN bracis-28371 319 34 rights right NOUN bracis-28371 319 35 and and CCONJ bracis-28371 319 36 permissions permission NOUN bracis-28371 319 37 reprints reprint NOUN bracis-28371 319 38 and and CCONJ bracis-28371 319 39 permissions permission VERB bracis-28371 319 40 copyright copyright NOUN bracis-28371 319 41 information information NOUN bracis-28371 319 42 © © ADP bracis-28371 319 43 2023 2023 NUM bracis-28371 319 44 the the DET bracis-28371 319 45 author(s author(s NOUN bracis-28371 319 46 ) ) PUNCT bracis-28371 319 47 , , PUNCT bracis-28371 319 48 under under ADP bracis-28371 319 49 exclusive exclusive ADJ bracis-28371 319 50 license license NOUN bracis-28371 319 51 to to ADP bracis-28371 319 52 springer springer NOUN bracis-28371 319 53 nature nature PROPN bracis-28371 319 54 switzerland switzerland PROPN bracis-28371 319 55 ag ag PROPN bracis-28371 319 56 about about ADP bracis-28371 319 57 this this DET bracis-28371 319 58 paper paper NOUN bracis-28371 319 59 cite cite VERB bracis-28371 319 60 this this DET bracis-28371 319 61 paper paper NOUN bracis-28371 319 62 kuhn kuhn PROPN bracis-28371 319 63 , , PUNCT bracis-28371 319 64 d.m d.m PROPN bracis-28371 319 65 . . PROPN bracis-28371 319 66 , , PUNCT bracis-28371 319 67 de de X bracis-28371 319 68 loreto loreto PROPN bracis-28371 319 69 , , PUNCT bracis-28371 319 70 m.s m.s PROPN bracis-28371 319 71 . . PROPN bracis-28371 319 72 , , PUNCT bracis-28371 319 73 recamonde recamonde NOUN bracis-28371 319 74 - - PUNCT bracis-28371 319 75 mendoza mendoza PROPN bracis-28371 319 76 , , PUNCT bracis-28371 319 77 m. m. NOUN bracis-28371 319 78 , , PUNCT bracis-28371 319 79 comba comba PROPN bracis-28371 319 80 , , PUNCT bracis-28371 319 81 j.l.d j.l.d NOUN bracis-28371 319 82 . . PROPN bracis-28371 319 83 , , PUNCT bracis-28371 319 84 moreira moreira PROPN bracis-28371 319 85 , , PUNCT bracis-28371 319 86 v.p v.p PROPN bracis-28371 319 87 . . PROPN bracis-28371 319 88 ( ( PUNCT bracis-28371 319 89 2023 2023 NUM bracis-28371 319 90 ) ) PUNCT bracis-28371 319 91 . . PUNCT bracis-28371 320 1 explainability explainability NOUN bracis-28371 320 2 of of ADP bracis-28371 320 3     SPACE bracis-28371 320 4 covid-19 covid-19 PROPN bracis-28371 320 5 classification classification NOUN bracis-28371 320 6 models model NOUN bracis-28371 320 7 using use VERB bracis-28371 320 8 dimensionality dimensionality NOUN bracis-28371 320 9 reduction reduction NOUN bracis-28371 320 10 of of ADP bracis-28371 320 11     SPACE bracis-28371 320 12 shap shap PROPN bracis-28371 320 13 values value NOUN bracis-28371 320 14 . . PUNCT bracis-28371 321 1 in in ADP bracis-28371 321 2 : : PUNCT bracis-28371 321 3 naldi naldi PROPN bracis-28371 321 4 , , PUNCT bracis-28371 321 5 m.c m.c PROPN bracis-28371 321 6 . . PROPN bracis-28371 321 7 , , PUNCT bracis-28371 321 8 bianchi bianchi PROPN bracis-28371 321 9 , , PUNCT bracis-28371 321 10 r.a.c r.a.c ADP bracis-28371 321 11 . . PUNCT bracis-28371 321 12 ( ( PUNCT bracis-28371 321 13 eds ed NOUN bracis-28371 321 14 ) ) PUNCT bracis-28371 321 15 intelligent intelligent ADJ bracis-28371 321 16 systems system NOUN bracis-28371 321 17 . . PUNCT bracis-28371 322 1 bracis bracis PROPN bracis-28371 322 2 2023 2023 NUM bracis-28371 322 3 . . PUNCT bracis-28371 323 1 lecture lecture NOUN bracis-28371 323 2 notes note NOUN bracis-28371 323 3 in in ADP bracis-28371 323 4 computer computer NOUN bracis-28371 323 5 science science NOUN bracis-28371 323 6 ( ( PUNCT bracis-28371 323 7 ) ) PUNCT bracis-28371 323 8 , , PUNCT bracis-28371 323 9 vol vol NOUN bracis-28371 323 10 14195 14195 NUM bracis-28371 323 11 . . PUNCT bracis-28371 324 1 springer springer NOUN bracis-28371 324 2 , , PUNCT bracis-28371 324 3 cham cham PROPN bracis-28371 324 4 . . PUNCT bracis-28371 325 1 https://doi.org/10.1007/978-3-031-45368-7_27 https://doi.org/10.1007/978-3-031-45368-7_27 PROPN bracis-28371 325 2 download download NOUN bracis-28371 325 3 citation citation NOUN bracis-28371 325 4 .ris .ris PUNCT bracis-28371 326 1 .enw .enw PROPN bracis-28371 326 2 .bib .bib PUNCT bracis-28371 327 1 doi doi PROPN bracis-28371 327 2 : : PUNCT bracis-28371 327 3 https://doi.org/10.1007/978-3-031-45368-7_27 https://doi.org/10.1007/978-3-031-45368-7_27 PROPN bracis-28371 327 4 published publish VERB bracis-28371 327 5 : : PUNCT bracis-28371 327 6 12 12 NUM bracis-28371 327 7 october october PROPN bracis-28371 327 8 2023 2023 NUM bracis-28371 327 9 publisher publisher NOUN bracis-28371 327 10 name name NOUN bracis-28371 327 11 : : PUNCT bracis-28371 327 12 springer springer NOUN bracis-28371 327 13 , , PUNCT bracis-28371 327 14 cham cham PROPN bracis-28371 327 15 print print PROPN bracis-28371 327 16 isbn isbn PROPN bracis-28371 327 17 : : PUNCT bracis-28371 327 18 978 978 NUM bracis-28371 327 19 - - SYM bracis-28371 327 20 3 3 NUM bracis-28371 327 21 - - PUNCT bracis-28371 327 22 031 031 NUM bracis-28371 327 23 - - PUNCT bracis-28371 327 24 45367 45367 NUM bracis-28371 327 25 - - PUNCT bracis-28371 327 26 0 0 NUM bracis-28371 327 27 online online ADJ bracis-28371 327 28 isbn isbn NOUN bracis-28371 327 29 : : PUNCT bracis-28371 327 30 978 978 NUM bracis-28371 327 31 - - SYM bracis-28371 327 32 3 3 NUM bracis-28371 327 33 - - PUNCT bracis-28371 327 34 031 031 NUM bracis-28371 327 35 - - PUNCT bracis-28371 327 36 45368 45368 NUM bracis-28371 327 37 - - PUNCT bracis-28371 327 38 7 7 NUM bracis-28371 327 39 ebook ebook NOUN bracis-28371 327 40 packages package NOUN bracis-28371 327 41 : : PUNCT bracis-28371 327 42 computer computer NOUN bracis-28371 327 43 sciencecomputer sciencecomputer NOUN bracis-28371 327 44 science science NOUN bracis-28371 327 45 ( ( PUNCT bracis-28371 327 46 r0 r0 NOUN bracis-28371 327 47 ) ) PUNCT bracis-28371 327 48 share share VERB bracis-28371 327 49 this this DET bracis-28371 327 50 paper paper NOUN bracis-28371 327 51 anyone anyone PRON bracis-28371 327 52 you you PRON bracis-28371 327 53 share share VERB bracis-28371 327 54 the the DET bracis-28371 327 55 following follow VERB bracis-28371 327 56 link link NOUN bracis-28371 327 57 with with ADP bracis-28371 327 58 will will AUX bracis-28371 327 59 be be AUX bracis-28371 327 60 able able ADJ bracis-28371 327 61 to to PART bracis-28371 327 62 read read VERB bracis-28371 327 63 this this DET bracis-28371 327 64 content content NOUN bracis-28371 327 65 : : PUNCT bracis-28371 327 66 get get VERB bracis-28371 327 67 shareable shareable ADJ bracis-28371 327 68 linksorry linksorry NOUN bracis-28371 327 69 , , PUNCT bracis-28371 327 70 a a DET bracis-28371 327 71 shareable shareable ADJ bracis-28371 327 72 link link NOUN bracis-28371 327 73 is be AUX bracis-28371 327 74 not not PART bracis-28371 327 75 currently currently ADV bracis-28371 327 76 available available ADJ bracis-28371 327 77 for for ADP bracis-28371 327 78 this this DET bracis-28371 327 79 article article NOUN bracis-28371 327 80 . . PUNCT bracis-28371 328 1 copy copy VERB bracis-28371 328 2 shareable shareable ADJ bracis-28371 328 3 link link NOUN bracis-28371 328 4 to to PART bracis-28371 328 5 clipboard clipboard NOUN bracis-28371 328 6 provided provide VERB bracis-28371 328 7 by by ADP bracis-28371 328 8 the the DET bracis-28371 328 9 springer springer NOUN bracis-28371 328 10 nature nature PROPN bracis-28371 328 11 sharedit sharedit PROPN bracis-28371 328 12 content content NOUN bracis-28371 328 13 - - PUNCT bracis-28371 328 14 sharing share VERB bracis-28371 328 15 initiative initiative NOUN bracis-28371 328 16 keywords keyword NOUN bracis-28371 328 17 mortality mortality NOUN bracis-28371 328 18 prediction prediction NOUN bracis-28371 328 19 classification classification NOUN bracis-28371 328 20 models model NOUN bracis-28371 328 21 explainable explainable ADJ bracis-28371 328 22 ai ai VERB bracis-28371 328 23 dimensionality dimensionality NOUN bracis-28371 328 24 reduction reduction NOUN bracis-28371 328 25 shap shap NOUN bracis-28371 328 26 values value NOUN bracis-28371 328 27 publish publish VERB bracis-28371 328 28 with with ADP bracis-28371 328 29 us us PROPN bracis-28371 328 30 policies policy NOUN bracis-28371 328 31 and and CCONJ bracis-28371 328 32 ethics ethic NOUN bracis-28371 328 33 search search NOUN bracis-28371 328 34 search search NOUN bracis-28371 328 35 by by ADP bracis-28371 328 36 keyword keyword NOUN bracis-28371 328 37 or or CCONJ bracis-28371 328 38 author author NOUN bracis-28371 328 39 search search NOUN bracis-28371 328 40 navigation navigation NOUN bracis-28371 328 41 find find VERB bracis-28371 328 42 a a DET bracis-28371 328 43 journal journal NOUN bracis-28371 328 44 publish publish VERB bracis-28371 328 45 with with ADP bracis-28371 328 46 us we PRON bracis-28371 328 47 track track VERB bracis-28371 328 48 your your PRON bracis-28371 328 49 research research NOUN bracis-28371 328 50 discover discover VERB bracis-28371 328 51 content content NOUN bracis-28371 328 52 journals journal NOUN bracis-28371 328 53 a a DET bracis-28371 328 54 - - PUNCT bracis-28371 328 55 z z NOUN bracis-28371 328 56 books book NOUN bracis-28371 328 57 a a DET bracis-28371 328 58 - - PUNCT bracis-28371 328 59 z z NOUN bracis-28371 328 60 publish publish NOUN bracis-28371 328 61 with with ADP bracis-28371 328 62 us us PROPN bracis-28371 328 63 journal journal PROPN bracis-28371 328 64 finder finder PROPN bracis-28371 328 65 publish publish VERB bracis-28371 328 66 your your PRON bracis-28371 328 67 research research NOUN bracis-28371 328 68 language language NOUN bracis-28371 328 69 editing edit VERB bracis-28371 328 70 open open ADJ bracis-28371 328 71 access access NOUN bracis-28371 328 72 publishing publishing NOUN bracis-28371 328 73 products product NOUN bracis-28371 328 74 and and CCONJ bracis-28371 328 75 services service NOUN bracis-28371 328 76 our our PRON bracis-28371 328 77 products product NOUN bracis-28371 328 78 librarians librarian VERB bracis-28371 328 79 societies society NOUN bracis-28371 328 80 partners partner NOUN bracis-28371 328 81 and and CCONJ bracis-28371 328 82 advertisers advertiser NOUN bracis-28371 328 83 our our PRON bracis-28371 328 84 brands brand NOUN bracis-28371 328 85 springer springer NOUN bracis-28371 328 86 nature nature PROPN bracis-28371 328 87 portfolio portfolio PROPN bracis-28371 328 88 bmc bmc PROPN bracis-28371 328 89 palgrave palgrave PROPN bracis-28371 328 90 macmillan macmillan PROPN bracis-28371 328 91 apress apress PROPN bracis-28371 328 92 discover discover VERB bracis-28371 328 93 your your PRON bracis-28371 328 94 privacy privacy NOUN bracis-28371 328 95 choices choice NOUN bracis-28371 328 96 / / SYM bracis-28371 328 97 manage manage NOUN bracis-28371 328 98 cookies cookie NOUN bracis-28371 328 99 your your PRON bracis-28371 328 100 us us PROPN bracis-28371 329 1 state state NOUN bracis-28371 329 2 privacy privacy NOUN bracis-28371 329 3 rights right NOUN bracis-28371 329 4 accessibility accessibility NOUN bracis-28371 329 5 statement statement NOUN bracis-28371 329 6 terms term NOUN bracis-28371 329 7 and and CCONJ bracis-28371 329 8 conditions condition NOUN bracis-28371 329 9 privacy privacy NOUN bracis-28371 329 10 policy policy NOUN bracis-28371 329 11 help help NOUN bracis-28371 329 12 and and CCONJ bracis-28371 329 13 support support VERB bracis-28371 329 14 legal legal ADJ bracis-28371 329 15 notice notice NOUN bracis-28371 329 16 cancel cancel VERB bracis-28371 329 17 contracts contract NOUN bracis-28371 329 18 here here ADV bracis-28371 329 19 129.74.145.123 129.74.145.123 NUM bracis-28371 329 20 hesburgh hesburgh PROPN bracis-28371 329 21 library library PROPN bracis-28371 329 22 er er INTJ bracis-28371 329 23 unit unit NOUN bracis-28371 329 24 ( ( PUNCT bracis-28371 329 25 3005732405 3005732405 NUM bracis-28371 329 26 ) ) PUNCT bracis-28371 329 27 northeast northeast ADJ bracis-28371 329 28 research research NOUN bracis-28371 329 29 libraries library NOUN bracis-28371 329 30 ( ( PUNCT bracis-28371 329 31 nerl nerl PROPN bracis-28371 329 32 ) ) PUNCT bracis-28371 329 33 ( ( PUNCT bracis-28371 329 34 8200828607 8200828607 NUM bracis-28371 329 35 ) ) PUNCT bracis-28371 329 36 nerl nerl VERB bracis-28371 329 37 ta ta X bracis-28371 329 38 account account NOUN bracis-28371 329 39 ( ( PUNCT bracis-28371 329 40 3006206169 3006206169 NUM bracis-28371 329 41 ) ) PUNCT bracis-28371 329 42 university university NOUN bracis-28371 329 43 of of ADP bracis-28371 329 44 notre notre PROPN bracis-28371 329 45 dame dame PROPN bracis-28371 329 46 hesburgh hesburgh PROPN bracis-28371 329 47 library library NOUN bracis-28371 329 48 ( ( PUNCT bracis-28371 329 49 3000184373 3000184373 NUM bracis-28371 329 50 ) ) PUNCT bracis-28371 330 1 © © ADP bracis-28371 330 2 2025 2025 NUM bracis-28371 330 3 springer springer NOUN bracis-28371 330 4 nature nature NOUN