TABLE 3.

ACC and AUC of ensemble models.

Classifier	Voting method	Validation set		Test set
		ACC (Mean ± Std)	AUC (Mean ± Std)	ACC	AUC
LR	—	0.7944 ± 0.0542	0.8607 ± 0.0547	0.7647	0.7885
SVM	—	0.7942 ± 0.0503	0.8634 ± 0.0517	0.7647	0.7885
RF	—	0.7744 ± 0.069	0.7815 ± 0.0741	0.7647	0.8269
XGBoost	—	0.7744 ± 0.078	0.7911 ± 0.1129	0.7647	0.8269
RF + XGBoost + LR	soft	0.7944 ± 0.0653	0.8465 ± 0.0659	0.8824	0.8654
XGBoost + SVM + LR	soft	0.8275 ± 0.0264	0.8632 ± 0.0559	0.8235	0.8462
RF + XGBoost + SVM	soft	0.8011 ± 0.0480	0.8453 ± 0.0684	0.8235	0.8654
RF + XGBoost + LR	hard	0.7811 ± 0.0695	—	0.8235	—
All	hard	0.8211 ± 0.0456	—	0.7647	—
all	soft	0.8144 ± 0.0275	0.8587 ± 0.0550	0.7647	0.8654

The best performance in the models is highlighted in bold.