Table 1 Different feature selection results
From: Prediction of neonatal deaths in NICUs: development and validation of machine learning models
Feature selection Method | Selected features |
|---|---|
Neonatologist opinion | 17 features BW, GA, Preterm birth, SGA, Parental care, Mother disease, RDS, Steroid therapy, Surfactant administration, Pulmonary hemorrhage, NEC, Congenital malformation, Sepsis, Asphyxia, IVH, Intubation, Ventilation |
Non-parametric Mann–Whitney and Chi-square test (p-value < 0.05) | 13 features BW, GA, Preterm birth, SGA, Mother disease, RDS, Surfactant administration, Pulmonary hemorrhage, Congenital malformation, Sepsis, IVH, Intubation, Ventilation |
Non-parametric Mann–Whitney and Chi-square test (p-value >  0.2) | 14 features BW, GA, Preterm birth, SGA, Mother disease, RDS, Surfactant administration, Pulmonary hemorrhage, Asphyxia, Congenital malformation, Sepsis, IVH, Intubation, Ventilation |
IBM SPSS modeler feature selection method | 12 features BW, GA, Preterm birth, SGA, RDS, Surfactant administration, Pulmonary hemorrhage, Congenital malformation, IVH, Intubation, Ventilation, NEC |
CfsSubsetEva method in Weka | 5 features SGA, Parental care, Pulmonary hemorrhage, Intubation, Ventilation |