Fig. 2

The efficiency of ML. Box plots, showing performance of each model and eight machine learning models (with and without oversampling) for predicting outcomes in test set. Horizontal line in box indicated median; ends of box indicate IQR; ends of whiskers indicate interdecile range; and points beyond whiskers indicate outliers. (A) AUC values of clinical models. (B) AUC values of PET models. (C) AUC values of clinical-PET models. (D) AUC values of Combined models.