Microsoft Word - BRAIN_vol_9_issue_2_2018_v4_final2_ok.doc 95 A Factor Analysis Model for Dimension Reduction of Outcome Factors in Neonatal Seizure Context Ionela Maniu Research and Telemedicine Center in Neurological Diseases in Children, Pediatric Clinical Hospital, Sibiu, Romania 3 George Barițiu, 550178, Tel. 0269 230 250 Department of Mathematics and Computer Science, “Lucian Blaga” University, Sibiu, Romania Bulevardul Victoriei 10, Sibiu 550024, Tel.: 0269 216 062 ionela.maniu@yahoo.ro George Maniu Department of Mathematics and Computer Science, “Lucian Blaga” University , Sibiu, Romania 10 Bulevardul Victoriei, 550024, Tel.: 0269 216 062 george.maniu@ulbsibiu.ro Cristina Dospinescu Research and Telemedicine Center in Neurological Diseases in Children, Pediatric Clinical Hospital, Sibiu, Romania 3 George Barițiu, 550178, Tel. 0269 230 250 cristinadospinescu@yahoo.com Gabriela Visa Research and Telemedicine Center in Neurological Diseases in Children, Pediatric Clinical Hospital, Sibiu, Romania 3 George Barițiu, 550178, Tel. 0269 230 250 gabiap@yahoo.com Abstract There is a controversial concept among many studies whether neonatal seizures are risk factors for neonatal death and/or neurodevelopment impairments (in case of newborn survivors). Multiple factors have been analyzed in literature, including perinatal factors, etiology factors, seizures characteristics factors, investigations findings factors, therapy-related factors. This paper aims to review the characteristics and the application context of different computational models developed for identifying both the risk factor of morbidity (epilepsy, cerebral palsy, development disability or their combination) and the mortality outcome after neonatal seizures. Consequently, we determined the groups of main risk factors using factor analysis. The vast majority of identified models are logistic regression models, but also decision tree models. In the literature, there is a large variation in establishing the risk factors determining poor or favorable outcome after a neonatal seizure, with similarities and inconsistencies. These findings could be a consequence of different approaches regarding inclusion criteria, methodologies used to identify seizure, seizures definition or description, analysis using computational models. Keywords: neonatal seizures, risk factors, epilepsy, mortality, logistic regression, decision tree 1. Introduction Presence of seizures and seizures control are very important aspects that clearly affect the intervention care phases and clinicians decision-making process. The differential diagnosis of neonatal and perinatal seizures is a complex task for a practitioner and it includes a lot of pathological conditions (epileptic and nonepileptic): 1. Nonepileptic (gastroesophageal reflux- Sandifer Syndrome, cyanotic breath-holding attacks, shuddering spells and jitteriness, pallid syncopal attacks-reflex anoxic attacks, hyperlexia, cardiac arrhythmias, subarachnoid hemorrhage, subdural hematoma, mitochondrial cytopathies, meningitis, encephalitis), 2.epileptic syndromes (benign, myoclonic, pyridoxine) (Cazan et. al., 2014; Cazan et. al., 2017). Hence, the differential BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 2 (May, 2018), ISSN 2067-3957 96 diagnosis algorithms should be undertaken carefully and efficiently in order to promptly intervene in the case management and to avoid any undesirable complication. Currently, there are many efforts to perform either noninvasive assessment of biomarkers in children with epilepsies or to apply different computational models to categorize the risk factors for susceptible cases (Shahar et. al., 2012; Choy, 2014; Azhibekov, 2015; Nicolae et. al., 2016; Pitkänen, 2016; Mahoney, 2016). Different outcomes from neonatal period affect not only the suffering patient but the whole community. Multiple prognostic factors have been analyzed in literature, including perinatal factors, etiology factors, seizures characteristics factors, investigations findings factors, therapy-related factors Costea et. al. 2017). This paper aims to review the characteristics of different computational models developed for identifying the risk factor of outcomes after neonatal seizures and to determine the groups of main risk factors. 2. Methods We examined NCBI databases (PMC, PubMed) using multiple search word combinations: neurological outcome, risk factors, neonatal seizure and neurological outcome, neonatal seizure, and epilepsy, neonatal seizure and mortality, selecting the articles including computational models analysis. The information from each article was synthesized in three synoptic tables. Table 1 presents the reviewed studies description: author, year of publication, study period, type. We discussed the study concepts linked with the seizures description or definition, the inclusion or exclusion criteria and with other methodologies used to identify seizure and different risk factors. Table 2 offers details concerning the characteristics of these models designed to identify the risk factors. Table 3 provides further information about the identified outcomes and risk factors using different computational models. Eventually, a factor analysis was implemented to define specific risk factors groups (table 4). Based on our previously presented work (Maniu et. al., 2017), we also considered in the factor analysis input (beside de reviewed studies from this paper) another five studies describing a different, scoring system based, approach (Ellison et. al., 1981, 1986; Pisani et. al., 2009; Garfinkle, & Shevell, 2011; Salamon et. al., 2014; Hur, & Chung, 2016). 3. Results We have found ten new highly related studies from the literature review, published between 2005 and 2016, of which two were multicenter studies (considering 2 and 4 centers) while the rest were one hospital based case series (Miller et. al., 2005; Ambalavanan, 2006; Nunes, 2008; Pisani, 2012; Yildiz, 2012; Lai, 2013; Vargas, 2013; Anand, 2014; Shah, 2014; Pisani, 2016). Their brief description is presented in table 1. Table 1. Reviewed studies description First author Year / Country Study period / Study type Study description Miller 2005 / California 1994-2000 R, HB, MC -173 term (GA>=36 wk) newborns with neonatal encephalopathy (2 centers, 121 (1994-2000) + 52) -gestational age median: 6 days (range 1-24 days) -magnetic resonance imaging (MRI) Ambalavanan 2006 /Maryland -/MC -205 neonates diagnosed as having hypoxic-ischemic encephalopathy - HT and control groups Nunes 2008 / Brazil 1999-2003 P (3659), PB -101 newborns with seizures -diagnosis of neonatal seizures was based on clinical observation -seizure etiology was based on positive clinical data, laboratory data and/or imaging studies (CUS, CT or MRI) Pisani -403 consecutive newborns with gestational age from 24 to 32 weeks I. Maniu, G. Maniu, C. Dospinescu, G. Visa - A Factor Analysis Model for Dimension Reduction of Outcome Factors in Neonatal Seizure Context 97 2012 / Italy 2000-2007 / R, HB (preterm) -clinical and EEG-confirmed neonatal seizures Yildiz 2012 / Turkey 2007-2009 / R,HB -112 newborns (ages of 23-44 months) after manifesting seizures in their first postnatal 28 days Lai 2013 / Taiwan 1999-2009 / R, HB -232 term infants with clinical neonatal seizure - 17 related risk factors were analyzed - clinical neonatal seizure, EEG, CUS Vargas 2013/Colombia 2008-2012 / CC, HB -GA < 44 wk with neonatal seizures, treated with phenobarbital -20 case (without response), 35 control (with adequate response) Anand 2014/India -/ R,HB -108 newborns with seizure - EEG, USG, CT, MRI Shah 2014 / UK 2007-2011 R,HB - 85 neonates from 4 centers undergoing 72 h of TH - study hypothesis: seizure burden is associated with cerebral tissue injury independent of amplitude-integrated EEG (aEEG) background activity Pisani 2016 / Italy 1999-2012 / R,HB - 76 preterm newborns with seizure - video-EEG confirmed seizure - CUS R retrospective, P prospective, CC case – control study, MC multicenter controlled trail, PB populational based, HB hospital based, C clinical, CT computed tomographic scan, MRI cerebral magnetic resonance imaging, CUS cranial ultrasonography / cerebral ultrasound, USG ultrasonography, EEG electroencephalogram (standard), CpH cord Ph, BpH blood Ph, HT therapeutic hypothermia It can be noticed that seizure diagnosis was based on clinical grounds and functional explorations naming neuroimaging and/or EEG procedures (conventional EEG, aEEG, vEEG, CUS, MRI). The minimum number of newborns considered in these studies was 55, while the maximum was 403 with a mean of 148 (SD=86.75, median=112, IQR: (98,175)) and a total of 2226 evaluated cases. Figure 1. Pediatric epilepsy in the context of a cerebral malformation (Research and Telemedicine Center in Neurological Diseases, Pediatric Clinical Hospital from Sibiu, Romania) The most frequently used computational model was the regression model, especially the binary logistic regression (table 2). The significance level considered varied significantly between studies (0.05-0.2). A few authors combined the significance level with OR (greater than 1). A few studies specified the model’s performance using specific indicators such as correct classification BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 2 (May, 2018), ISSN 2067-3957 98 rates, sensitivity, and specificity. The performance values varied from 67% to 80% in case of death or disability and from 71% to 77% in case of death or 85.5% (in overall outcomes of the situation). Table 2 Characteristics of different computational models developed for identifying the risk factors Study Computational models Reported indicators Fisher exact test for qualitative (binary and non binary) variables N, %, p Kruskal-Wallis tests, Spearman rank correlation, bootstrap modeling for non-normality investigation for numeric variables Median (range), p Univariate linear regression score change, 95% CI, p Miller 2005 Multivariate model score change, 95% CI, p early neurologic examination correct classification rates logistic regression OR scoring system correct classification rates Ambalavanan 2006 classification and regression tree analysis correct classification rates Fisher and chi square tests for qualitative (only binary) variables N, %, RR, 95% CI, p Student’s t test for numeric variables M, SD, 95% CI, p Nunes 2008 multiple logistic regression model β coefficient, t, p Pisani 2012 Multivariate analysis OR, 95% CI, p Yildiz 2012 Multivariate logistic regression OR, 95% CI Fisher and chi square tests for categorized data N, %, p Lai 2013 simple and multiple logistic regression models N, %, OR, 95% CI, p contingency tables, Fisher and chi square tests Vargas 2013 logistic regression M, SD OR, 95% CI, p Anand 2014 Chi-square test N, %, p Shah 2014 Univariate and multivariate logistic regresion OR, 95% CI, p Student’s t test for numeric variables chi square tests for qualitative variables Pisani 2016 multivariate logistic regression model N,% sensitivity, specificity Outcomes like death, brain injury, cerebral palsy, developmental delay and/or epilepsy were displayed in table 3 for further analysis in our review. In some studies, different comorbidities were individually evaluated determining risk factors for each outcome. In other reports they were considered together (overall outcomes). Only in three research papers, treatment factors were considered and identified as risk factors. Table 3 Risk factors of outcomes and mortality in neonatal seizures Study Outcomes Identified risk factors Miller 2005 Brain injury BW, intensive resuscitation, severe encephalopathy, severe seizure Ambalavanan 2006 death or moderate/severe disability at 18 to 22 months or death as the outcomes few components of the early neurologic examination were associated with poor outcomes epilepsy (EP) abnormal PNN, abnormal PEEG, BW Multivariate analysis – 0 factors Nunes 2008 developmental delay abnormal NN, abnormal PNN, abnormal PEEG, earlier PNS, GA, BW I. Maniu, G. Maniu, C. Dospinescu, G. Visa - A Factor Analysis Model for Dimension Reduction of Outcome Factors in Neonatal Seizure Context 99 Multivariate analysis: abnormal PN, abnormal PEEG mortality GA Multivariate analysis – 0 factors Pisani 2012 mortality brain damage abnormal UBS, BW (<1000), cardiopulmonary resuscitation Yildiz 2012 cerebral palsy, epilepsy, developmental delay etiology, Apgar score, resuscitation, electroencephalogram, neonatal status epilepticus, cranial imaging findings, type/duration of antiepileptic treatment, response to acute treatment Lai 2013 death, cerebral palsy, global developmental delay, and/or epilepsy abnormal UBS, abnormal cerebral artery resistance index, abnormal EEG, presence of congenital heart disease Vargas 2013 risk for therapeutic failure with phenobarbital and posible feature complications antecedents at birth/adaptation: seizure semiology (more than one), latency to phenobarbital startup greater than 12 hours, subtle or tonic seizures, seizures longer than 5 minutes postnatal antecedents: low AS10, maternal intrapartum infection, neonatal shock Anand 2014 Overall seizure outcome Low GA, low BW, low AS5, etiology, SO, ST_EPI, abnormal radiological findings, abnormal EEG Shah 2014 Overall seizure outcome electrographic seizure burden Pisani 2016 Overall seizure outcome (death, development delay, cerebral palsy, epilepsy) BW, AS1, neurologic exam, EEG, UBS, presence of ST_EPI AED - antiepileptic drug, CP - cerebral palsy, GDD - global developmental delay, GA - gestational age, BW- birth weight, RS - repeated/recurrent seizure, MD - type/mode of delivery, AS1 - Apgar score at 1 minute, AS5 - Apgar score at 5 minute, AS10 - Apgar score at 10 minute, SO - seizure onset, ST_EPI - status epilepticus, UBS - ultrasound brain scan, MSU- maternal substance used, MIS – maternal inflammatory state, PRM- prolonged rupture of membranes, PNN – postnatal neuroimaging, PNS – postnatal seizure. The most frequently identified risk factors were the EEG findings (abnormal / severe electroencephalogram results), seizure characteristics (type, onset, duration, semiology), etiology, birth weight, Apgar score, cerebral ultrasound scan findings (abnormal) (Figure 2). Figure 2. Identified risk factors hierarchy BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 2 (May, 2018), ISSN 2067-3957 100 As can be seen from Table 3, many different factors have been identified in previous studies as risk factors in neonatal seizures. In order to group these large number of factors, factor analysis was used. Factor analysis’ main objective is to reduce many items into fewer latent factors by grouping similar variables into dimensions on the bases of pattern correlations. We consider reducing the risk factors dimension using factor analysis with principal axis factoring extraction method and varimax with Kaiser Normalization rotation method in our attempt to identify latent risk factors. The principal component analysis finds for the initial data a new orthonormal basis having the axes ordered depending on variance (from high to low). The new ordered vectors (eigenvectors) are the principal components. The method generated 5 main factors, this number being extracted from the examination of scree plot and eigenvalues (over 1). First latent factor gathered the treatment and the resuscitation records, the second one included Apgar score and status epilepticus. Delivery mode, gestational age, and birth weight (medical history factors) were grouped as the third-factor category. The forth category comprised the etiology and seizure characteristics while the fifth considered functional and imaging data (EEG, ultrasound brain scans, neuroimaging -MRI, CT) (table 4). Table 4. Factor analysis reduction factors rotated matrix Factor 1 2 3 4 5 treatment duration .892 response to treatment .738 .510 resuscitation .715 early neurologic examination -.482 status epilepticus .937 AS .905 mode of delivery .847 BW -.631 .440 GA .627 etiology .863 SO .669 ultrasound brain scan -.454 -.404 EEG -.439 .703 neuroimaging .702 4. Discussions and conclusions Prominently, the inclusion and exclusion criteria, the methodologies used to identify seizures, the description or the definition of seizures were different in these studies. Anand et. al. (2014) referred in their study to the clinical diagnosed seizures, newborns lacking the synchronized video EEG recordings. The affected newborns with very subtle or electrical only seizures might have been excluded. Lai (2012) had a similar approach analzying the patients with clinically evident seizures. On the other hand, Pisani et. al. (2016) studied only the newborns with confirmed video- EEG. A more complex analysis was presented in Shah 2014 study, where the aEEG and MRI were considered for diagnosis. Seizures pattern recorded on aEEG with corresponding 2-channel raw EEG (aEEG/EEG), were classified by severity of background and seizure burden; MR images were interpreted based on the severity of tissue injury. Different risk factors related to the medical history were considered by different studies. Lai, 2014 considered 17 variables such as gender, delivery mode, small for date status, maternal illness, perinatal insults, meconium stained liquor, Apgar score at 1 and 5 minutes, seizure onset age, seizure type, etiology, electroencephalography (EEG) findings, antiepileptic drug efficacy, presence of metabolic acidosis, cranial ultrasonographic findings, and the presence of congenital heart disease. Conversely, Anand et. al. (2014) selected the following variables: time of onset of seizure, type, duration and frequency of seizure, neurological examination at the onset of the seizure, gestational age, type of delivery, birth weight, Apgar score at 1 and 5 min, resuscitation at the time I. Maniu, G. Maniu, C. Dospinescu, G. Visa - A Factor Analysis Model for Dimension Reduction of Outcome Factors in Neonatal Seizure Context 101 of birth. Biomarker assessment for these patients however such as routine chemistries including blood sugar, sepsis screen, lumbar puncture, serum electrolytes such as sodium, calcium, magnesium, EEG, neuroimaging (ultrasonography/computerised tomography/magnetic resonance imaging [USG/CT/MRI]), TORCH screening and inborn errors of metabolism screening were performed whenever indicated by the clinician and we couldn’t recognize a systematic approach in this respect. In contrast, Ambalavanan used clinical and the laboratory variables (available within 6 hours of birth) to develop the model (both scoring system and decision tree). As a consequence, distinctive computational models and distinctive approaches in using the models were employed in these reports with a predictable variability in the identified risk factors. In some analysis, univariate and multivariate regression techniques were applied only on categorical variables while in other studies there were both categorical and numerical variables. We didn’t find any description of the categories selection. Moreover, we noticed differences in cut of points considered for the same variable (risk factor) between authors. Vargas et. al. (2013) considered that there was a higher risk with more than six medical antecedents at birth or with more than five postnatal medical antecedents. Miller et. al. (2005) findings pointed out that the measured prenatal risk factors did not predict the brain injury pattern types as neuroimaging biomarkers (watershed predominant, basal ganglia/thalamus predominant, and normal) but that the patterns of brain injury in terms of neonatal encephalopathy are associated with different clinical presentations and neurodevelopmental outcomes. The large amount of factors identified in previous studies as risk factors for outcomes in neonatal seizures context make their analysis difficult to accomplish. We consider our approach to be an interesting and practical design for future computational models. It operates five major groups that cluster the most important studied risk factors on the reviewed studies. By using this type of data reduction technique, the large number of factors considered as risk factors for outcome in neonatal seizure context are restructured and we can identify derive groups of factors / underlying factors to be considered by the specialists in intervention care phases but also in prognostic and prediction of clinical outcome. Although there are controversial discussions regarding the use of factor procedure in case of categorical (binary) variables, the resulted groups have clinical relevance. Nonetheless, there are other limitations for this model related to the number of the included studies. Further work on metadata and considering also other modeling techniques (principal component analysis for categorical variables procedures (tetrachoric correlation), association rules) are needed to validate and optimize the risk factors groups. Acknowledgment This work has been conducted in the Pediatric Clinic Hospital Sibiu, within Research and Telemedicine Center in Neurological Diseases in Children - CEFORATEN project (ID 928 SMIS- CSNR 13605) financed by ANCSI with the grant number 432 / 21. 12. 2012 thru the Sectoral Operational Programme „Increase of Economic Competitiveness”. References Ambalavanan N, Carlo WA, Shankaran S, Bann CM, Emrich SL, Higgins RD, Tyson JE, OShea TM, Laptook AR, Ehrenkranz RA, Donovan EF, Walsh MC, Goldberg RN, Das A; National Institute of Child Health and Human Development Neonatal Research Network. Predicting outcomes of neonates diagnosed with hypoxemic-ischemic encephalopathy. Pediatrics 2006;118(5):2084-2093. Anand V, Nair P. Neonatal seizures: Predictors of adverse outcome. J Pediatr Neurosci 2014;9:97-9 Azhibekov T, Soleymani S, Lee BH, Noori S, Seri I. Hemodynamic monitoring of the critically ill neonate: An eye on the future. Seminars in Fetal & Neonatal Medicine. 2015;20(4):246– 254. BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 2 (May, 2018), ISSN 2067-3957 102 Cazan C, Dobrota L, Neamtu M PO-0813 Sandifer Syndrome – A Challenge For The Paediatrician Archives of Disease in Childhood 2014; 99:A520. http://dx.doi.org/10.1136/archdischild- 2014-307384.1447 Cazan C, Dobrota L, Neamtu M, Bogdan Neamtu, Gabriela Visa, Nicolae Bodrug, Viorel Istrate Elevated aminotransferase activity as a clue to muscular dystrophy Archives of Disease in Childhood 2017;102:A148. http://dx.doi.org/10.1136/archdischild-2017-313273.385 Choy M, Dube CM, Patterson K, Barnes SR, Maras P, Blood AB, Hasso AN, Obenaus A, Baram TZ. A novel, noninvasive, predictive epilepsy biomarker with clinical potential. Journal of Neuroscience 2014; 34: 8672-8684 Costea R., Neamtu M, Neamtu B. (2016) Febrile seizures – EEG study and risks factors for recurrence and for epilepsy. European Journal of Pediatrics, 2016, ISSN: 175:1393-1880; 1432-1076; pp 1616; DOI 10.1007/s00431-016-2785-8 Ellison PH, Largent JA, Bahr JP. A scoring system to predict outcome following neonatal seizures. J Pediatr [Internet]. 1981;99(3):455–9. Available from: http://ovidsp.ovid.com/ovidweb.cgi?T=JS&PAGE=reference&D=med2&NEWS=N&AN=7 264808 Ellison PH, Horn JL, Franklin S, Jones MG. The results of checking a scoring system for neonatal seizures. Neuropediatrics [Internet]. 1986;17(3):152–7. Available from: http://ovidsp.ovid.com/ovidweb.cgi?T=JS&CSC=Y&NEWS=N&PAGE=fulltext&D=med2 &AN=2429227%5Cnhttp://openurl.ac.uk/athens:_edu//lfp/LinkFinderPlus/Display?sid=OV ID:Ovid+MEDLINE(R)&id=pmid:2429227&id=&issn=0174- 304X&isbn=&volume=17&issue=3&spage=152&pages=152-7& Garfinkle J, Shevell MI. Prognostic factors and development of a scoring system for outcome of neonatal seizures in term infants. Eur J Paediatr Neurol [Internet]. Elsevier Ltd; 2011;15(3):222–9. Available from: http://dx.doi.org/10.1016/j.ejpn.2010.11.002 Hur J.Y, Chung ML. Risk Factor Analysis and Scoring System for Neurodevelopmental Outcomes after Neonatal Seizures. J Neonatal Biol [Internet]. 2016;5(4). Available from: https://www.omicsgroup.org/journals/risk-factor-analysis-and-scoring-system-for- neurodevelopmentaloutcomes-after-neonatal-seizures-2167-0897-1000241.php?aid=83335 Lai YH, Ho CS, Chiu NC, Tseng CF, Huang YL. Prognostic factors of developmental outcome in neonatal seizures in term infants. Pediatr Neonatol 2013; 54: 166-72 Mahoney, L. G. S. (2016) Assessment of the transitional circulation in late preterm and term neonates using non-invasive biomarkers: a longitudinal analysis and evaluation of repeatability (Doctoral dissertation, University of Brighton) Maniu I, Maniu G.C, Balan C, Neamtu B, Scoring systems for predicting outcomes of neonatal seizure, Acta Medica Transilvanica (in press) Miller, S. P., Ramaswamy, V., Michelson, D., Barkovich, A. J., Holshouser, B., Wycliffe, N., … Ferriero, D. M. (2005). Patterns of brain injury in term neonatal encephalopathy. J Pediatr, 146(4), 453–460. https://doi.org/10.1016/j.jpeds.2004.12.026, http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=5EBB0E03036AAFB375605D053 46F1A5E?doi=10.1.1.328.3903&rep=rep1&type=pdf Neamtu B, Picu O, Martu Stefanache MA, Ibric Cioranu VS (2016) The Comparative Evaluation of Salivary Biomarkers (Calcium, Phosphate, Salivary pH) in Cario-resistance Versus Cario- activity Vasile Nicolae, REV.CHIM.(Bucharest), 67, No. 4 , 2016 http://www.revistadechimie.ro Nunes ML, Martins MP, Barea BM, Wainberg RC, da Costa JC. Neurological outcome of newborns with neonatal seizures. Arq Neuropsiquiatr. 2008;66:168–174 Pisani F, Sisti L, Seri S. A Scoring System for Early Prognostic Assessment After Neonatal Seizures. Pediatrics [Internet]. 2009;124(4):e580–7. Available from: http://pediatrics.aappublications.org/cgi/doi/10.1542/peds.2008-2087 Pisani F, Piccolo B, Cantalupo G, et al. Neonatal seizures and post-neonatal epilepsy: a seven-year follow-up study. Pediatr Res 2012;72:186e93 I. Maniu, G. Maniu, C. Dospinescu, G. Visa - A Factor Analysis Model for Dimension Reduction of Outcome Factors in Neonatal Seizure Context 103 Pisani F, Facini C, Pelosi A, Mazzotta S, Spagnoli C, Pavlidis E (2016) Neonatal seizures in preterm newborns: A predictive model of outcome. Eur J Paediatr Neurol 20:243-251 Pitkänen A, Löscher W, Vezzani A, et al. Biomarkers in epilepsy. Lancet Neurol 2016;15:843–856 Salamon S.A, Neubauer D, Petrovcic A, Paro-Panjan D. Risk factors and scoring system as a prognostic tool for epilepsy after neonatal seizures. Pediatr Neurol [Internet]. Elsevier Ltd; 2014;50(1):77–84. Available from: http://dx.doi.org/10.1016/j.pediatrneurol.2013.08.010 Shahar E, Nagler R. (2012) Oxidative Stress in Children with Intractable Epilepsy Reflected by Salivary Glands Anti-Oxidant Profile , Neurology , 78 (1 Supplement) P01.073; ISSN:1526- 632X Shah DK, Wusthoff CJ, Clarke P, et al , Electrographic seizures are associated with brain injury in newborns undergoing therapeutic hypothermia, Archives of Disease in Childhood - Fetal and Neonatal Edition 2014;99:F219-F224 Vargas N.P, Sánchez E.F, Segura O.E (2014) RISK FACTORS ASSOCIATED WITH A POOR RESPONSE TO FIRST-LINE TREATMENT IN NEONATAL SEIZURES A CASE- CONTROL STUDY DONE AT MERCY FOUNDATION HOSPITAL, Revista Academia Nacional de Medicina, Vol 103, https://encolombia.com/medicina/revistas- medicas/academedicina/vol103/tratamiento-crisis-neonatales/ Yıldız EP, Tatlı B, Ekici B, et al. Evaluation of etiologic and prognostic factors in neonatal convulsions. Pediatr Neurol 2012; 47: 186-92 Ionela Maniu is a lecturer professor Ph.D at Lucian Blaga University of Sibiu, Faculty of Sciences, Department of Mathematics and Computer Science and researcher in the Neurology Research Department within the Research and Telemedicine Center in Neurological Diseases in Children affiliated to the Pediatric Clinical Hospital from Sibiu Romania. Current research is focused on data mining, machine learning and artificial intelligence techniques, with special interest in medical data sets.