Table 9.

AUC, accuracy, sensitivity, and specificity of each ML algorithm.

ML Models	5-Fold Cross-Validation				10-Fold Cross-Validation
	AUC	Accuracy	Sensitivity	Specificity	AUC	Accuracy	Sensitivity	Specificity
Control vs. PSD
SVM_L	0.690	0.600	0.748	0.557	0.706	0.636	0.775	0.542
SVM_R	0.708	0.646	0.681	0.495	0.711	0.700	0.742	0.517
KNN	0.659	0.538	0.743	0.352	0.681	0.579	0.742	0.425
RF	0.685	0.538	0.619	0.557	0.696	0.560	0.767	0.600
VE	0.675	0.615	0.676	0.552	0.646	0.650	0.708	0.517
Imp vs. NoImp
SVM_L	0.830	0.771	0.600	0.883	0.797	0.775	0.650	0.950
SVM_R	0.496	0.648	0.267	0.817	0.722	0.708	0.300	0.800
KNN	0.635	0.681	0.300	0.950	0.674	0.742	0.300	0.850
RF	0.760	0.743	0.467	0.867	0.624	0.717	0.500	0.950
VE	0.784	0.743	0.533	0.867	0.747	0.733	0.450	0.90

Abbreviations: PSD = poststroke depression; Imp = PSD patients showing improvement in their symptoms; NoImp = PSD patients showing no symptom improvement; SVM_L = linear support vector machine; SVM_R = support vector machine with radial basis function kernel function; KNN = k-nearest neighbors; RF = random forest; VE = voting ensemble; AUC = area under the curve.