2.9. Lasso cox regression model construction and validation

To find the genes related to prognosis among the differentially expressed genes and construct a prognostic model to provide indications for clinical treatment, we constructed a Lasso Cox regression model. First, we used the differentially expressed genes between different immunophenotypes to analyse the significance of single‐gene prognosis (P < 0.001). Subsequently, Cox regression analysis was performed using a single‐gene prognostic analysis of significant genes prognosis combined with survival information. To further optimize the model, we combined with the Lasso regression model to build a multi‐factor prognostic prediction model using the level of risk to divide patients into high‐risk and low‐risk groups and analysed the prognosis of the two groups to prove the effectiveness of the model. To verify the effectiveness of the model, additional data ({"type":"entrez-geo","attrs":{"text":"GSE31684","term_id":"31684"}}GSE31684) were used to validate the constructed model based on TCGA‐BLCA. The ‘survminer’, ‘survival’ and ‘ggplot’ packages were used for survival analysis.