Leveraging Interpretable Models and Low Complexity Models for Early Breast Cancer Diagnosis: A Machine Learning Approach

A crucial component of women’s healthcare is the identification of breast cancer, which necessitates precise and understandable predictive models. Even though machine learning has great potential, obstacles including unbalanced datasets, computing complexity, and interpretability impede advancement. To overcome these obstacles, we used a novel strategy in this study that concentrated on lightweight and interpretable models. In particular, we use decision trees, K-Nearest Neighbors (K-NN), and Support Vector Machines (SVM) for breast cancer diagnosis using logistic regression as a meta-learner. Our study uses the Wisconsin Breast Cancer (WBC) dataset, a gold standard in breast cancer research, to demonstrate the efficacy of this ensemble technique. Through the utilization of base models’ simplicity and the interpretability of logistic regression, we are able to achieve diagnosis transparency and accuracy, which helps physicians make well-informed decisions.