Table 3. Comparison of the predictive models in the validation cohort.

Models	RI model (imaging features alone)	RC model (clinical factors alone)	RIC model (imaging & clinical factors)
Sensitivity (95% CI)	60.0% (32.3–83.7%)	60.0% (32.3–83.7%)	66.7% (38.4–88.2%)
Specificity (95% CI)	97.3% (93.3–99.3%)	98.0% (94.2–99.6%)	98.7% (95.2–99.8%)
AUC (95% CI)	0.784 (0.713–0.844)	0.685 (0.608–0.755)	0.895* (0.838–0.938)

*, ROC comparison by Delong’s test: RIC model versus RI model, P=0.204, z statistic =1.270; RIC model versus RC model, P=0.015, z statistic =2.421; RI model versus RC model, P=0.446, z statistic =0.763. RI, radiomics imaging; RC, radiomics clinical; RIC, radiomics imaging and clinical; ROC, receiver operating characteristic; AUC, area under the curve.