Development of a machine learning-based model for predicting risk of early postoperative recurrence of hepatocellular carcinoma

Figure 5 Receiver operating characteristic analyses and validation of six prediction models. A: Receiver operating characteristic (ROC) curves of the six prediction models in the training dataset; B: ROC curves of the six prediction models in the validation dataset after five-fold cross-validation; C: Calibration plots of nine models. The XGBoost achieved lower (better) Brier scores than the other models; D: Decision curve analysis of six machine learning models. The XGBoost model is the best diagnostic tool for early postoperative hepatocellular carcinoma recurrence. ROC: Receiver operating characteristic.