Optimizing Breast Cancer Prognosis with Machine Learning for Enhanced Clinical DecisionMaking

Objective: Breast cancer remains a leading cause of mortality among women worldwide. Early detection and accurate prognosis are crucial for improving patient outcomes. This study presents a novel approach that integrates feature elimination techniques with machine learning to enhance the accuracy of breast cancer prognosis. The approach addresses class imbalance in the dataset to improve sensitivity, particularly in minimizing false negatives. Additionally, it emphasizes the use of machine learning algorithms, which are considered more transparent and computationally efficient compared to deep learning methods. Method: The Wisconsin Breast Cancer (WBC) dataset was used to develop an interpretable machine learning model. Recursive Feature Elimination (RFE) identified key features, while Principal Component Analysis (PCA) reduced dimensionality. The optimized feature set was trained using XGBoost. To address class imbalance, class weighting and decision threshold adjustments were applied to improve sensitivity and minimize false negatives. Results: The model achieved high performance: accuracy of 99.12%, precision of 100%, recall of 97.69%, and an F1 score of 98.9%. Feature selection and class imbalance handling enhanced sensitivity and computational efficiency. The model's interpretable results highlight its suitability for clinical applications. Conclusions: This study presents an interpretable machine learning model integrating RFE, PCA, and XGBoost to enhance breast cancer prognosis. High accuracy and sensitivity, coupled with explainability, make it a promising tool for clinical decisionmaking in early detection and treatment planning.