Fig. 8.

Construction of a diagnostic model based on core targets of BPs exposure-induced tumorigenesis in KIRC. A Performance evaluation of six ensemble machine learning models. ROC-AUC: Area under the curve of the receiver operating characteristic. B SHAP feature contribution ranking of core targets across six models. C Recursive feature selection based on AUC values. AUC: Area under the curve. D Bar plot of variable importance score based on the Gradient Boosting Machine. E Receiver Operating Characteristic (ROC) curves of the diagnostic model on training and validation datasets