Table 2.

Evaluation metrics of the models constructed by each algorithm

	AUC	ACC	SEN	SPE	Kappa	Brier score	MCC
Train
CART	0.981	0.893	1	0.883	0.577	0.038	0.637
GBM	0.965	0.94	0.843	0.949	0.684	0.038	0.694
KNN	0.969	0.957	0.941	0.959	0.777	0.047	0.788
LR	0.784	0.703	0.843	0.689	0.228	0.075	0.319
NNet	0.849	0.831	0.765	0.838	0.37	0.065	0.42
RF	0.978	0.924	0.941	0.922	0.651	0.041	0.682
XGBoost	0.996	0.959	1	0.955	0.794	0.017	0.881
Valid
CART	0.997	0.963	1	0.959	0.822	0.023	0.835
GBM	0.991	0.938	1	0.931	0.729	0.029	0.757
KNN	0.983	0.975	0.917	0.982	0.866	0.051	0.867
LR	0.75	0.529	0.917	0.486	0.133	0.081	0.242
NNet	0.907	0.793	0.917	0.78	0.376	0.058	0.459
RF	0.99	0.905	1	0.894	0.627	0.042	0.676
XGBoost	0.998	0.971	1	0.968	0.857	0.016	0.866

Train Training set

Valid Validation set

AUC Area under the curve

ACC Accuracy

SEN Sensitivity

SPE Specificity

MCC Matthews correlation coefficient