TABLE 2.

Performance of different models in training and test cohort.

Model_group	AUC	AUC_CI_low	AUC_CI_high	Sensitivity	Specificity	PPV	NPV	Accuracy
Clinical_train	0.676	0.545	0.79	0.467	0.774	0.538	0.719	0.663
Clinical_test	0.644	0.4	0.848	0.375	0.654	0.25	0.773	0.588
Image_train	0.661	0.519	0.802	0.154	0.965	0.667	0.714	0.711
Image_test	0.579	0.356	0.827	0.4	0.917	0.667	0.786	0.765
Combined_train	0.97	0.937	0.997	0.769	0.965	0.909	0.902	0.904
Combined_test	0.908	0.783	1	0.9	0.917	0.818	0.957	0.912

The combined model shows superior performance across all metrics compared to clinical or image models alone, with notably higher AUC, sensitivity, and accuracy in both cohorts.