Table 4. Sensitivity, Specificity and overall accuracy in identifying patients with a SDI≥1 (A) and SDI≥5 (B) by artificial neural networks analysis and traditional statistics.

	n° of variables	SN (%)	SP (%)	Accuracy (%)	ROC AUC
		(95% CI)	(95% CI)	(95% CI)	(95%CI)
A
ANNs	17	72.5 (65.91–79.09)	78.5 (72.75–84.00)	75.5 (71.13–79.87)	0.714** (0.673–0.755)
LR	45	35.8 (30.22–44.38)	76.5 (70.40–83.61)	56.2 (51.78–60.62)	0.616 (0.576–0.656)
B
ANNs	25	74.8 (62.89–80.72)	87.8 (83.22–92.38)	81.3 (76.44–86.16)	0.823** (0.780–0.866)
LR	45	37.3 (25.14–49.46)	90.3 (86.16–94.44)	63.8 (57.88–69.72)	0.699 (0.657–0.741)

ANNs: artificial neural networks; LR: logistic regression analysis; SN: sensitivity; SP: specificity; ROC: receiver operating characteristic; AUC: area under the curve.

**: p<0.01.