2.1. Establishment and Verification of Gene Prognostic Model

We carried out the LASSO Cox regression model (R package “glmnet”) to select the candidate genes and subsequently established the prognostic model [14, 15]. We ultimately retained 17 genes and their coefficients and utilized the minimum criteria to determine the penalty parameter (λ). We calculated the risk score after centralizing and standardizing (applying the “scale” function in R) the TCGA expression data. The risk score formula was shown as follows: risk score = ∑7iXi × Yi (X: coefficients, Y: gene expression level).