Skip to main content
. 2022 Jan 6;20:583–597. doi: 10.1016/j.csbj.2022.01.003

Fig. 5.

Fig. 5

Random survival forest accurately predicts test cohort outcomes. (A) Schematic representation of model testing strategy. (B) Predicted per-patient EFS probabilities over time for each patient in the test cohort. (C) Time-dependent ROC-AUC for random survival forest test set performance. (D) Prevalence and values of top 7 variables in each test cohort patient. Patients were clustered by event status and time to event/censoring. (E) Kaplan-Meier curve for patients with top-7 variable sums>0.5 (n = 15) versus less than or equal to 0.5 (n = 32).