Table 2.

Discriminative performance of different predictive models in the training and validation cohorts.

Predictive models	Training cohort				Validation cohort
	AUC (95% CI)	Accuracy	Sensitivity	Specificity	AUC (95% CI)	Accuracy	Sensitivity	Specificity
Clinical-radiological model	0.744 (0.642 - 0.846)	0.682	0.512	0.841	0.757 (0.595 - 0.920)	0.757	0.667	0.842
Radiomics model
AP	0.774 (0.675 - 0.873)	0.682	0.659	0.705	0.752 (0.592 - 0.911)	0.649	0.667	0.632
PVP	0.797 (0.705 - 0.890)	0.682	0.610	0.750	0.830 (0.684 - 0.977)	0.784	0.833	0.737
DP	0.736 (0.629 - 0.843)	0.682	0.561	0.795	0.757 (0.592 - 0.923)	0.730	0.667	0.789
AP-PVP	0.818 (0.729 - 0.907)	0.718	0.683	0.750	0.810 (0.671 - 0.949)	0.757	0.667	0.842
AP-DP	0.780 (0.681 - 0.879)	0.718	0.732	0.705	0.804 (0.652 - 0.956)	0.703	0.833	0.579
PVP-DP	0.800 (0.707 - 0.893)	0.706	0.683	0.727	0.830 (0.690 - 0.971)	0.757	0.889	0.632
AP-PVP-DP	0.838 (0.753 - 0.922)	0.753	0.732	0.773	0.833 (0.691 - 0.975)	0.703	0.889	0.526
Combined model	0.878 (0.806 - 0.950)	0.812	0.805	0.818	0.833 (0.687 - 0.979)	0.730	0.833	0.632

AP, arterial phase; PVP, portal venous phase; DP, delayed phase; AUC, area under the curve; CI, confidence interval.