Clinical Decision Support Meets Machine Learning: Statistical Evaluation of Optimized Feature Selection

Abstract

Breast cancer (BC) is one of the most prevalent diseases and ranks as the second leading cause of mortality in both developed and developing nations.. BC is a serious problem and causes a considerable mortality rate worldwide for women. The diagnosis of BC includes differentiation between malignant and benign tumors, accurate differentiation is crucial. As a result, automating this classification process is essential to reduce reliance on a physician’s experience and subjective judgment. The primary objective of this study is to apply multiple machine learning techniques to classify tumors as malignant or benign using the Wisconsin Breast Cancer Diagnostic dataset and to determine the most accurate classifier among them.. The performance comparison of classifiers with different feature selection methods is summarized as follows. We conclude that Gradient Boosting and XGBoost provide robust results, but their performance does not surpass that of simpler models like Logistic Regression.