FIGURE 5.

Feature selection and base model construction and comparison. (A) Feature selection: the SVM-REF and varSelRF algorithms were used for feature selection, and 29 features were retained. (B) Base model construction and comparison: nine machine learning algorithms were used to fit models based on 5-fold cross-validation (CV) on the training cohort (TCGA). The performance of each model on CV was comprehensively evaluated.