Table 4. Comparison of the prediction performances of the prediction models on the testing dataset (n=209,860)

Model	Accuracy	Specificity	Sensitivity	Balanced accuracy	AUROC
The final double-ensemble model	0.6933	0.6933	0.691	0.6922	0.7538
GBM + LGBM	0.6529	0.6523	0.7004	0.6764	0.7421
GBM + LR	0.6925	0.6926	0.6845	0.6886	0.7530
LGBM + LR	0.6791	0.6788	0.7004	0.6896	0.7533
GBM + LGBM + LR	0.6851	0.6851	0.6853	0.6852	0.7513

Abbreviations: AUROC, area under receiver of characteristics; XGB, XGBoost; GBM, gradient boosting machine; LGBM, light gradient-boosting machine; AdaBoost, adaptive boosting; LR, logistic regression.