Figure 2

Verification of the stability of the prognosis prediction model including 14 immune-related genes of GBM patients in the training set. (A) The 1–3-year overall survival (OS) predicted receiver operating characteristic (ROC) curves of a 14-gene risk model in the training set. (B) The distribution of samples in Risk-H and Risk-L groups of the training set was done using the 14-gene risk model under different OS. (C) The level of Risk-L group/Total sample size with the extension in OS in the training set. (D) The clustering results of the training set samples. Fourteen genes were used for hierarchical clustering. The distance between different features was calculated by a Euclidean distance analysis. These genes clustered into high- and low-expression groups, and samples in the training set were also divided into two groups. (E) Difference in the RiskScore between the two groups, which had been clustered by the expression of 14 genes of training set samples.