A Hybrid Approach To Accurate Breast Cancer Prediction Integrating: Explainable AI and Machine Learning

Female breast cancer functions as a leading mortality cause of cancer among women globally. Discovery of breast cancer at its early stage enhances both therapeutic outcomes and patient survival rates. Although traditional diagnostic techniques function well they require extended time along with high costs and show limitations in human error rates. The project establishes an automatic breast cancer detection system through machine learning algorithms whereDecision Tree and Support Vector Machine (SVM) function as supervised learning engines. This system applies the Breast Cancer Wisconsin Diagnostic Dataset that contains 30 numerical features extracted from digitized images of breast masses from FNA samples to reach its evaluation. The features consist of radius, texture, perimeter, area, smoothness together with concavity measurements that describe the cell nuclei characteristics. The development process includes data normalization along with scaling and separates data into training and testing portions for model implementation and evaluation. The model includes two classifiers: The Decision Tree as an understandable solution and SVM which excels at processing complex features. The evaluation metrics for both classifiers include standard measures that encompass accuracy together with precision and recall and F1-score. Explainable AI (XAI) methods have been integrated to make machine learning models more transparent because medical professionals require such methods to reinforce trust. The SHAP (SHapley Additive exPlanations) interpretive method shows global feature importance as well as local prediction importance so doctors can fully understand feature effects on model outputs. LIME provides individual prediction explanations through local model approximation with interpretable surrogates.