Table 5.

Predictive performance of each model for HCC PD-L1 expression.

Cohort	Prediction model	AUC	Sen	Spe	PPV	NPV	ACC	F1-score
Training cohort	FS-T2WI model	0.731	0.391	0.902	0.642	0.767	0.743	0.486
	Arterial phase model	0.800	0.435	0.922	0.714	0.783	0.770	0.541
	Portal venous phase model	0.800	0.435	0.882	0.625	0.776	0.743	0.513
	Multisequence model	0.831	0.565	0.902	0.722	0.821	0.797	0.634
	Radiomics-clinical model	0.898	0.652	0.922	0.790	0.855	0.838	0.714
Validation cohort	FS-T2WI model	0.621	0.272	0.870	0.500	0.714	0.676	0.353
	Arterial phase model	0.743	0.545	0.956	0.857	0.815	0.824	0.667
	Portal venous phase model	0.771	0.364	0.957	0.800	0.759	0.765	0.500
	Multisequence model	0.810	0.454	1.000	1.000	0.793	0.824	0.625
	Radiomics-clinical model	0.779	0.636	1.000	1.000	0.852	0.882	0.778

AUC, area under the receiver operating characteristic curve; ACC, accuracy; Sen, sensitivity; Spe, specificity; PPV, positive predictive value; and NPV, negative predictive value.