Enhanced Multi-step Breast Cancer Prediction Through Integrated Dimensionality Reduction and Support Vector Classification

Cancer is a dreaded disease which can affect any body part. Breast cancer is a significant health alarm worldwide, especially distressing millions of women each year. It is a notable pathological condition with high mortality rates especially in the developing nations. Traditional machine learning models have been working for breast cancer prediction, but they face challenges when dealing with high-dimensional and noisy data. Notably, early detection can lead to timely detection and eradication of the associated mortality risk. This approach can help battle the associated concern and stigma with breast cancer in a big way. This manuscript proposes an enhanced multi-step approach for breast cancer prediction by integrating dimensionality reduction with the SVM classifier. This integrated approach not only improves the prediction accuracy but also enhances interpretability by identifying the most relevant features associated with breast cancer. When applied to the WDBC dataset the approach gave 96.50% accuracy.