Explainable AI for Breast Cancer Diagnosis: Comparative Analysis of ML Models Using Random Forest Feature Selection and SHAP Interpretability

Abstract

Breast cancer diagnosis is critical for improving patient outcomes, yet traditional methods face limitations such as invasiveness and human error. This study presents an explainable AI framework for breast cancer classification using six ML models: LR, NB, KNN, RF, SVC, and DT. SMOTE addresses class imbalance, while RF feature selection reduces dimensionality from 30 to 19 features. SHAP interpretability is integrated to provide clinical insights into feature contributions, enhancing trust in model predictions. The SVC model with RF-selected features achieves superior performance, with an accuracy of 0.9930 and recall of 1.0000, highlighting the importance of features such as smoothness mean. This framework balances accuracy, efficiency, and transparency, offering a foundation for clinical deployment and guiding future work on external validation and broader adoption of explainable ML in breast cancer care.