Table 2.

Results of the models based on enhanced CT imaging

Model	Metrics	Balanced_accuracy	AUC	Specificity	Sensitivity	F1 score
Radiomics model	Training set	0.689 (0.644–0.724)	0.857 (0.804–0.900)	0.406 (0.340–0.468)	0.971 (0.928–1.000)	0.573 (0.504–0.634)
	Internal validation set	0.634 (0.556–0.712)	0.781 (0.680–0.875)	0.402 (0.297–0.505)	0.867 (0.741–0.970)	0.556 (0.444–0.652)
	External validation set	0.665 (0.515–0.813)	0.711 (0.524–0.872)	0.565 (0.360–0.773)	0.765 (0.556–0.944)	0.65 (0.457–0.810)
Clinical model	Training set	0.656 (0.597–0.715)	0.786 (0.597–0.715))	0.941 (0.907–0.971)	0.371 (0.258–0.486)	0.872 (0.838–0.903)
	Internal validation set	0.737 (0.646–0.824)	0.836 (0.646–0.824)	0.908 (0.844–0.958)	0.567 (0.390–0.739)	0.883 (0.830–0.926)
	External validation set	0.662 (0.523–0.794)	0.733 (0.523–0.794)	0.913 (0.773–1.000)	0.412 (0.188–0.650)	0.778 (0.640–0.885)
Clinincal-radiomics model	Training set	0.794 (0.729–0.851)	0.910 (0.729–0.851)	0.931 (0.896–0.963)	0.657 (0.529–0.763)	0.908 (0.876–0.936)
	Internal validation set	0.832 (0.747–0.911)	0.876 (0.747–0.911)	0.931 (0.872–0.978)	0.733 (0.565–0.885)	0.920 (0.878–0.959)
	External validation set	0.684 (0.555–0.802)	0.821 (0.555–0.802)	0.957 (0.850–1.000)	0.412 (0.182–0.636)	0.800 (0.654–0.897)