Fig. 3.

Risk model performance in the testing and entire TCGA data. Distribution of survival status and survival times for individuals in testing set (A) and entire set (F). Risk score for each one in the testing dataset (B) and whole dataset (G). Kaplan-Meier survival curves between low- and high-risk group in the testing (C) and entire datasets (H). ROC evaluation of 1-, 3-, 5-year survival prediction in the testing (D) and entire datasets (I). Clinical ROC curve showing the prediction accuracy between the risk score and every clinical feature at one year in the testing (E) and whole datasets (J).