Machine learning-based analysis for prediction of surgical necrotizing enterocolitis in very low birth weight infants using perinatal factors: a nationwide cohort study

Table 1 Performance metrics of the proposed model for prediction of sNEC

	Sensitivity	Specificity	PLR	NLR	PPV	NPV	F1-score	AUROC	Global accuracy	Post-test probability
Linear SVM	0.6812	0.6508	1.9911	0.4938	0.6905	0.6667	0.6771	0.6850	0.6812	0.5413
Radial SVM	0.6913	0.6430	1.9933	0.4864	0.6905	0.6698	0.6840	0.6736	0.6913	0.5410
Logistic regression	0.6857	0.6514	1.9943	0.4849	0.6894	0.6652	0.6822	0.6836	0.6857	0.5434
KNN	0.5012	0.6011	1.2938	0.8373	0.5963	0.5428	0.4544	0.6042	0.5012	0.4310
XGBOOST	0.6314	0.6353	1.7597	0.5833	0.6580	0.6186	0.6280	0.6461	0.6314	0.5118
LightGBM	0.6406	0.6489	1.8517	0.5557	0.6762	0.6364	0.6328	0.6534	0.6406	0.5246
Random forest	0.6656	0.6644	2.0607	0.5098	0.6925	0.6552	0.6597	0.6869	0.6656	0.5467
Proposed model	0.7049	0.6496	2.0297	0.4564	0.7010	0.6867	0.6983	0.7210	0.7049	0.5486

Pre-test probability is 0.6019
PLR positive likelihood ratio, NLR negative likelihood ratio, PPV positive predictive value, NPV negative predictive value, SVM support vector machine, KNN K-nearest neighbors, XGBOOST extreme gradient boosting, LightGBM light gradient-boosting machine