Fig. 6

Machine learning to construct stemness models. A LASSO, Boruta, SVM and XGBoost feature selection performance evaluation, AUC is generated by ROC curve analysis. B The feature genes shared by the four machine learning algorithms were identified by VENN plots, totaling 22 important feature genes. C ROC curves of 22 gene signatures predicted to validate stemness subtypes. D Heat map showing the expression of the signature genes in the validation set, with red indicating high expression and blue indicating low expression. The top of the heat map shows the distribution of clinical features for each patient, including Stemness Subtype, Age, Gender, Histology, Stage, Stage, Status, Relapse. E Survival differences between Stemness Subtype I and Stemness Subtype II groups. Red indicates Stemness Subtype I and green indicates Stemness Subtype II