Table 3.

Performance of classifiers using 115 (originally tagged) and 19 (automatically selected) features.

Classifier and feature sets		Sensitivity, mean (SD)	Specificity, mean (SD)	AUCa, mean (SD)
GNBb
	All 115 features	0.685 (0.125)	0.771 (0.116)	0.822 (0.062)
	Automatically selected 19 features	0.634 (0.074)	0.903 (0.063)	0.882 (0.054)
KNNc
	All 115 features	0.973 (0.013)	0.526 (0.096)	0.901 (0.032)
	Automatically selected 19 features	0.943 (0.028)	0.703 (0.048)	0.935 (0.023)
XGBd
	All 115 features	0.982 (0.01)	0.766 (0.059)	0.978 (0.012)
	Automatically selected 19 features	0.970 (0.019)	0.737 (0.051)	0.970 (0.016)

^aAUC: area under the receiver operating characteristic curve.

^bGNB: Gaussian Naive Bayes.

^cKNN: K-nearest neighbor algorithm.

^dXGB: extreme gradient boosting.