Table 3.

Predictive performance of Models 1 and 2 in the training and test set.

	Training set (n = 116)				Test set (n = 52)
	AUC (95% CI)	ACC	SEN	SPE	AUC (95% CI)	ACC	SEN	SPE
Model 1	0.749 (0.702–0.804)	0.724	0.706	0.738	0.739 (0.665–0.818)	0.692	0.652	0.724
Model 2	0.800 (0.759–0.845)	0.776	0.725	0.815	0.783 (0.718–0.860)	0.808	0.826	0.793

Model 1 included independent clinicopathological markers and Model 2 included both radiomic features and clinicopathological markers.

AUC, area under the curve; ACC, accuracy; CI, confidence interval; SEN, sensitivity; SPE, specificity.