Skip to main content
. 2026 Aug 7;19:621065. doi: 10.2147/IJGM.S621065

Figure 4.

Two line graphs showing receiver operating characteristic curves and calibration curves for prediction models. Image A shows a ROC graph with False Positive Rate (0.0-1.0) on the x-axis and True Positive Rate (0.0-1.0) on the y-axis. Legend: Clinical only (AUC 0.707), Clinical + body composition (AUC 0.806), Clinical + radiomics (AUC 0.738), Clinical + body composition + radiomics (AUC 0.830), BISAP (AUC 0.448), EPIC (AUC 0.545), mPASS (AUC 0.627). Curves rise from (0.0, 0.0) to (1.0, 1.0). Image B shows a calibration graph with Mean predicted probability (0.0-1.0) on the x-axis and Fraction of positives (0.0-1.0) on the y-axis. Legend: Perfectly calibrated, Clinical only, Clinical + body composition, Clinical + radiomics, Clinical + body composition + radiomics. Lower right text: Clinical only (Brier 0.225), Clinical + body composition (Brier 0.192), Clinical + radiomics (Brier 0.212), Clinical + body composition + radiomics (Brier 0.171).

Receiver operating characteristic curves and calibration curves based on five-fold out-of-fold predictions. (A) ROC curves compare the clinical-only, clinical + body composition, clinical + radiomics, and clinical + body composition + radiomics models, together with conventional clinical scoring systems. (B) Calibration curves show the agreement between predicted and observed probabilities for the four XGBoost-based models.