Fig. 3

Distribution of risk score (high or low) and status (dead or alive) and KM curves of OS in the training (left), testing (medium), and combined (right) sets