Table 1.

Evaluation of the classifier performance.

FS strategy	TNR	TPR	PPV	F0.5
Semi-automatic	0.96 ± .01	0.41 ± .04	0.77 ± .05	0.62 ± .04
Fully automatic	0.96 ± .01	0.41 ± .04	0.77 ± .05	0.63 ± .04

Estimated classification scores with 95% confidence intervals computed through averaging on a cross-validation process based on two feature selection strategies (FSS). TNR true negative rate, TPR true positive rate, PPV positive predictive value.