Figure 3

MRS model validation. (A–C) The training dataset was analyzed using Kaplan–Meier (A), ROC (B), and Time-ROC curves (C). (D–F) The validation dataset was analyzed using Kaplan–Meier (D), ROC (E), and Time–ROC curves (F). Univariate and multivariate Cox regression analyses were performed for MRS values in the training (G) and validation datasets (H). An MRS-based nomogram was constructed for the training (I) and validation datasets (J), with corresponding 1-, 3-, and 5-year calibration curves for these nomograms in the training (K) and validation datasets (L). *P < 0.05; **P < 0.01; ***P < 0.001.