Fig. 5

Development of the Macro.RF Derived Subtype System for Prognosis and Treatment Efficacy Prediction in BC. A Flowchart illustrating the construction of the Macro.RF model using machine learning processes. B Comparison of multiple ROC curves representing the performance of different machine learning algorithms in the validation set. C Schematic representation of molecular subtypes based on M_Macrophage-SPP1-C1Q. D ROC curve depicting the performance of the final Macro.RF model in validation and test cohorts. E Kaplan–Meier curves comparing OS between patients in the LM (good prognosis) and HM (poor prognosis) groups in the validation and test sets. F, G Clinical response rates to chemotherapy in patients stratified into LM (good response) and HM (poor response) groups across different cohorts