Table 9.

The performance of the best ML models developed using different sets of top features on the validation dataset. Features were selected using mean-based univariate analysis.

Total feature	ML model	Sensitivity	Specificity	Accuracy	AUC	Kappa	MCC
Top 305	RF	57.63	66.67	60.87	0.72	0.22	0.23
Top 200	RF	66.10	69.7	67.39	0.73	0.34	0.34
Top 150	RF	59.32	72.73	64.13	0.73	0.29	0.31
Top 100	ET	67.80	66.67	67.39	0.73	0.33	0.33
Top 50	ET	61.02	66.67	63.04	0.70	0.26	0.27
Top 20	SVC	50.85	75.76	59.78	0.70	0.23	0.26
Top 10	SVC	59.32	69.70	63.04	0.69	0.27	0.28

RF, random forest; ET, extra tree; SVC, support vector classifier.