Table 3.

Performance of the individualized prediction models

Models	Train cohort
	AUC	95% CI	Sensitivity	Specificity	Accuracy
T	0.716	0.631~0.802	0.480	0.816	0.660
N	0.783	0.700~0.866	0.750	0.716	0.732
AJCC	0.776	0.701~0.852	0.846	0.666	0.750
C	0.810	0.731~0.889	0.846	0.716	0.776
A	0.734	0.642~0.827	0.826	0.600	0.705
V	0.694	0.596~0.793	0.634	0.733	0.687
A + V	0.795	0.711~0.879	0.730	0.750	0.741
A + C	0.844	0.772~0.916	0.788	0.800	0.794
V + C	0.826	0.751~0.901	0.807	0.766	0.785
A + V + C	0.856	0.788~0.924	0.865	0.716	0.785
Models	Test cohort
	AUC	95% CI	Sensitivity	Specificity	Accuracy
T	0.661	0.515~0.806	0.437	0.750	0.645
N	0.764	0.624~0.905	0.812	0.625	0.687
AJCC	0.696	0.565~0.827	0.812	0.562	0.645
C	0.764	0.621~0.908	0.812	0.656	0.708
A	0.708	0.540~0.877	0.750	0.468	0.562
V	0.603	0.429~0.777	0.625	0.406	0.479
A + V	0.705	0.530~0.879	0.687	0.656	0.666
A + C	0.791	0.636~0.945	0.750	0.687	0.708
V + C	0.785	0.642~0.928	0.750	0.656	0.687
A + V + C	0.792	0.635~0.950	0.750	0.593	0.645

Note: A, V and A + V indicate the predicted model based on arterial phase images, venous phase images and the combination of two phase images, respectively. T, N and AJCC indicate the predicted model based on T stage, N stage and AJCC stage, respectively. C indicates the predicted model based on the combination of T stage, N stage and AJCC stage. CI Confidence Interval