Table 3.

Performance on the training set of the three machine learning techniques using a 5-fold cross-validation method.

Technique	Sensitivity, mean (95% CI)	PPVa, mean (95% CI)	NPVb, mean (95% CI)	Specificity, mean (95% CI)	F score, mean (95% CI)
GLMNetc	80.2 (77.7-82.7)	73.2 (70.9-75.6)	90.9 (89.6-92.2)	87.1 (85.6-88.7)	76.5 (75.6-77.5)
MAXENTd	68.8 (66.8-70.7)	66.0 (62.5-69.5)	86.1 (85.2-86.9)	84.5 (82.7-86.3)	67.4 (64.7-70.0)
Boosting	86.6 (82.1-91.1)	95.8 (93.2-98.5)	94.4 (92.4-96.3)	98.3 (97.0-99.6)	90.9 (89.7-92.1)

^aPPV: positive predicative value.

^bNPV: negative predicative value.

^cGLMNet: elastic-net regularized generalized linear model.

^dMAXENT: maximum entropy.