Table 2

Performance of prediction models for predicting the response to LPI therapy in patients with unresectable HCC in the training and validation cohorts

Cohort	Model	AUC (95% CI)	ACC	SEN	SPE	PPV	NPV	Youden index
Training cohort	Clinical model	0.669 (0.571 to 0.766)	0.644	0.608	0.680	0.660	0.630	0.288
	Radiomics model	0.900 (0.842 to 0.958)	0.832	0.686	0.980	0.972	0.754	0.666
	Combined model	0.912 (0.860 to 0.964)	0.842	0.706	0.980	0.973	0.766	0.686
Validation cohort	Clinical model	0.585 (0.442 to 0.728)	0.580	0.714	0.483	0.500	0.700	0.197
	Radiomics model	0.893 (0.804 to 0.982)	0.820	0.952	0.724	0.714	0.955	0.677
	Combined model	0.892 (0.803 to 0.981)	0.840	0.810	0.862	0.810	0.862	0.672

ACC, accuracy; AUC, area under the curve; HCC, hepatocellular carcinoma; LPI, lenvatinib plus PD-1 inhibitors and interventional; NPV, negative predictive value; PPV, positive predictive value; SEN, sensitivity; SPE, specificity.