Table 7.

Performances of SVMs selected using different metrics on validation set for 3 cases with various sizes of training sets

	SVM selected using:	Score [95% CI]			Difference from Ref. $\overline{{\mathcal{L}}_w}$ *
		AUC	Bacc	$\overline{{\mathcal{L}}_w}$
Case 1	AUC	0.91 [0.88 0.93]	0.83 [0.78 0.87]	−3.55 [−5.30 −2.02]	48.89%
	Bacc	0.84 [0.78 0.89]	0.75 [0.70 0.80]	−3.71 [−5.68 −2.00]	55.60%
Case 2	AUC	0.91 [0.88 0.94]	0.83 [0.79 0.87]	−3.26 [−4.91 −1.83]	36.75%
	Bacc	0.85 [0.80 0.90]	0.76 [0.71 0.81]	−3.51 [−5.37 −1.93]	47.35%
Case 3	AUC	0.91 [0.88 0.94]	0.84 [0.80 0.88]	−2.49 [−3.86 −1.31]	4.35%
	Bacc	0.91 [0.88 0.94]	0.83 [0.79 0.87]	−2.86 [−4.37 −1.53]	20.06%

¹ Reference $\overline{{\mathcal{L}}_w}$ =−2.39 [−3.26, −1.58]