Table 2.

The sensitivity and specificity each of the studies using different classifiers and hold-out set sizes in leave-$n$-out cross-validation obtained, along with their 95% confidence interval. It also shows statistical analysis including likelihood ratio, the area under the receiver operator characteristics curve and its confidence interval, and the $p$-value showing the likelihood of obtaining each result.

Classifier, hold-out set size	Sensitivity (%)	95% CI (%)		Specificity (%)	95% CI (%)		+LR	C-Stat	95% CI		$p$-values
Steatosis
SVM, 2	74	[68	79]	72	[67	77]	2.66	0.71	[0.67	0.74]	$<0.00010$
SVM, 10	71	[64	77]	73	[68	78]	2.61	0.70	[0.66	0.74]	$<0.0001$
RF, 2	40	[33	48]	86	[82	89]	2.90	0.67	[0.62	0.71]	$<0.0001$
RF, 10	47	[39	55]	85	[82	88]	3.19	0.69	[0.64	0.73]	$<0.0001$
Adenomyosis
SVM, 1	82	[64	92]	74	[61	84]	3.14	0.77	[0.69	0.86]	$<0.0001$
RF, 1	58	[40	74]	78	[66	86]	2.63	0.69	[0.59	0.79]	0.0012
Craniosynostosis
SVM, 1	95	[72	99]	82	[62	93]	5.25	0.89	[0.80	0.98]	0.0006
RF, 1	82	[62	93]	77	[46	93]	3.60	0.80	[0.67	0.92]	0.0037