Fig. 4

Evaluation of the models. A Cumulative distribution curves of the Rad scores. The ROCs for the training (B) and the test (C) cohorts. The calibration curves demonstrated good calibration of the models in both cohorts (D–G)