TABLE 4.

Comparison of different models for the three optimal feature subsets on 10-fold cross-validation.

Feature Subset	Model Name	10-fold CV
		AUC	ACC (%)	SEN (%)	SPE (%)	MCC
DS + RDKit_237	BRF	0.8446	76.77	82.36	74.40	0.5263
	EEC	0.8406	77.11	81.34	75.31	0.5239
	BBC + XGBoost	0.8248	78.10	73.26	80.16	0.5151
	BBC + GBDT	0.8438	80.08	75.78	81.90	0.5557
	BBC + LightGBM	0.8295	79.58	75.91	81.14	0.5490
	BBC + SVM	0.8089	79.25	64.33	85.60	0.5055
	BBC + MLP	0.8185	76.61	77.17	76.37	0.5040
	BBC + KNN	0.8281	74.64	78.86	72.85	0.4772
	BBC + LR	0.7850	71.52	69.05	72.57	0.3926
MOE + RDKit_196	BRF	0.8429	76.28	83.85	73.07	0.5252
	EEC	0.7806	71.99	73.27	71.45	0.4150
	BBC + XGBoost	0.8058	76.28	75.11	76.78	0.4932
	BBC + GBDT	0.8041	78.43	73.09	80.69	0.5183
	BBC + LightGBM	0.8194	77.11	72.62	79.01	0.4965
	BBC + SVM	0.7869	76.43	68.69	79.72	0.4679
	BBC + MLP	0.8019	75.95	74.24	76.68	0.4839
	BBC + KNN	0.8318	74.62	78.59	72.93	0.4732
	BBC + LR	0.7666	69.70	72.68	68.43	0.3859
SubFPC	BRF	0.8459	76.10	83.06	73.14	0.5224
	EEC	0.7614	69.20	77.27	65.76	0.3963
	BBC + XGBoost	0.8061	75.29	70.76	77.21	0.4547
	BBC + GBDT	0.8416	78.75	73.71	80.89	0.5256
	BBC + LightGBM	0.8237	76.78	74.28	77.84	0.4961
	BBC + SVM	0.7764	73.96	68.31	76.36	0.4271
	BBC + MLP	0.7909	75.80	68.64	78.84	0.4552
	BBC + KNN	0.7963	73.79	71.38	74.82	0.4336
	BBC + LR	0.7554	69.72	68.10	70.41	0.3588