Figure 1

Kaplan-Meier survival curves of HCC patients in the training set. Relative hazard (h) was predicted for cancer survival (A) and disease-free survival (B) using a leave-one-out procedure. Patients were equally divided into two groups based on h, and their Kaplan-Meier survival functions were compared by log-rank test. Alternatively, we divided patients into three equal-sized groups based on their h, and observed excellent separation. Such results suggest the prediction is rather robust, and not sensitive to choice of grouping. The vertical bars on the survival curve denote censored patients.