Fig. 2

The process of feature selection. (A) The correlation coefficients of the baseline characteristics with distant metastasis. (B) The heatmap of Spearman’s correlation analysis among the clinical variables and distant metastasis. The correlation index ranges from -1.0 to 1.0, with a brighter color indicating a stronger correlation. (C) The forest plot visualized the feature selection with multivariate logistic regression analysis. (D-I) Feature selection process with Recursive Feature Elimination (RFE) method based on six ML algorithms (GBM, SVM, RF, DT, XGB and CatBoost).