Table 2. Performance of the machine learning algorithms.

Algorithm	Oversampling method	Area under ROC curve	Matthews correlation coefficient	Brier score	Sensitivity	Specificity	Accuracy
Logistic regression	SMOTE#	0.830	0.433	0.036	0.692	0.968	0.965
	ADASYN*	0.823	0.376	0.049	0.692	0.955	0.968
Support vector machine	SMOTE#	0.825	0.393	0.045	0.692	0.959	0.970
	ADASYN*	0.786	0.345	0.048	0.615	0.958	0.971
K nearest neighbor	SMOTE#	0.644	0.253	0.031	0.307	0.981	0.942
	ADASYN*	0.759	0.410	0.028	0.538	0.979	0.924
Random forest	SMOTE#	0.787	0.351	0.046	0.615	0.959	0.972
	ADASYN*	0.787	0.351	0.046	0.615	0.959	0.971
Gradient boosting	SMOTE#	0.787	0.351	0.046	0.615	0.959	0.971
	ADASYN*	0.787	0.351	0.046	0.615	0.959	0.971

Notes:

^#SMOTE, Synthetic minor oversampling technique.

^*ADASYN, Adaptive synthetic sampling.