Table 2.

Prediction performance of radscore, deepscore, and radiomics nomogram models in the training and validation sets.

Model	AUC(CI)	Sensitivity	Specificity	PPV	NPV	Youden Index
Train set
Tra_rad	0.874(0.802-0.946)	0.764	0.900	0.913	0.735	0.664
Deep_rad	0.778(0.677-0.879)	0.836	0.675	0.780	0.750	0.511
Final_rad	0.923(0.865-0.980)	0.964	0.800	0.869	0.941	0.764
Test set
Tra_rad	0.765( 0.678-0.853)	0.941	0.590	0.667	0.920	0.531
Deep_rad	0.652( 0.544-0.759)	0.765	0.538	0.591	0.724	0.303
Final_rad	0.842(0.762-0.923)	0.941	0.744	0.762	0.935	0.685