Table 2.

Predictive performances of the top-3 models and their fusion on the training/validation and independent testing sets.

Classifier + feature selection	Training/Validation set (n = 88)				Testing set (n = 23)
	AUC	ACC	SEN	SPE	AUC	ACC	SEN	SPE
Randomforest + trace_ratio	0.788	0.704	0.70	0.67	0.792	0.739	0.818	0.667
Randomforest + f_score	0.776	0.684	0.65	0.65	0.78	0.70	0.727	0.667
k-Nearest Neighbor + f_score	0.775	0.70	0.784	0.61	0.803	0.70	0.818	0.583
Ensemble Methods	PV				0.795	0.739	0.818	0.667
	WF				0.811	0.783	0.818	0.75

The highest values are marked in bold.