Table 4.

Diagnostic performance of the including models.

Model	Combined sensitivity (95%CI)	Combined specificity (95%CI)	Combined positive LR (95%CI)	Combine NegativeLR (95%CI)	Combined DOR (95%CI)	AUROC (95%CI)
Alexnet	0.79(0.66, 0.88)	0.80(0.65, 0.90)	4(1.9, 8.5)	0.26(0.14, 0.50)	15(4, 61)	0.86(0.83–0.89)
Densenet	0.87(0.83, 0.91)	0.87(0.80, 0.92)	6.6(4.1, 10.7)	0.15(0.11, 0.21)	45(22, 95)	0.93(0.91, 0.95)
Efficientnet	0.88(0.84, 0.9)	0.73(0.66, 0.78)	3.2(2.5, 4)	0.17(0.13, 0.23)	19(12, 30)	0.9(0.87, 0.92)
Inception	0.81(0.61 ,0.92)	0.86(0.78, 0.92)	5.8(4.0, 8.4)	0.23(0.11, 0.47)	26(14, 49)	0.9(0.88, 0.93)
Mobilenet	0.89(0.87, 0.9)	0.86(0.82, 0.89)	6.3(4.9, 8.0)	0.13(0.12, 0.16)	47(34, 64)	0.9(0.87, 0.93)
Resnet	0.91(0.87, 0.94)	0.9(0.86, 0.93)	8.9(6.2, 12.9)	0.1(0.06, 0.16)	89(43, 187)	0.96(0.94, 0.97)
VGG	0.87(0.82, 0.92)	0.86(0.79, 0.91)	6.4(4.1, 10)	0.15(0.1, 0.22)	44(20, 94)	0.93(0.91, 0.95)
TOTAL	0.87(0.85, 0.89)	0.85(0.82, 0.87)	6.67(5.74, 7.76)	0.14(0.12, 0.16)	49(38, 65)	0.94(0.91, 0.96)

This table is shown to present the models’ information of the diagnostic performance. After the statistical analysis, the statistics are shown in the table. The combined sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio and AUROC of each model is explained.

AUROC = area under the receiver operating characteristic curve, CI = confidence interval; DOR = diagnostic odds ratio; positive LR = positive likelihood ratio; negative LR = negative likelihood ratio.