Table 5.

Leave one out cross-validation (LOOCV) for the traditional neuropsychological tests [A], the indices of EXIT-360° and the traditional neuropsychological tests [B], and the indices of EXIT-360° [C].

Methods	AUC	CA	F1	Precision	Recall
[A] Traditional neuropsychological tests
k-nearest neighbors (kNN)	0.67	0.61	0.61	0.61	0.61
Logistic regression	0.67	0.61	0.61	0.61	0.61
Naive bayes	0.68	0.65	0.65	0.65	0.65
Support vector machine (SVM)	0.68	0.64	0.63	0.63	0.64
[B] EXIT-360° and Traditional Neuropsychological tests
k-nearest neighbors (kNN)	0.86	0.80	0.80	0.80	0.80
Logistic regression	0.93	0.90	0.90	0.90	0.90
Naive bayes	0.85	0.80	0.80	0.80	0.80
Support vector machine (SVM)	0.90	0.81	0.81	0.81	0.81
[C] EXIT-360°
k-nearest neighbors (kNN)	0.86	0.79	0.79	0.79	0.79
Logistic regression	0.91	0.85	0.85	0.85	0.85
Naive bayes	0.91	0.83	0.83	0.83	0.83
Support vector machine (SVM)	0.91	0.85	0.85	0.86	0.85

AUC (Area under the ROC curve) is the area under the classic receiver-operating curve; CA (Classification accuracy) represents the proportion of the examples that were classified correctly; F1 represents the weighted harmonic average of the precision and recall (defined below); Precision represents a proportion of true positives among all the instances classified as positive. In our case, the proportion of conditions correctly identified; Recall represents the proportion of true positives among the positive instances in our data.