EMPOWERING EARLY BREAST CANCER DETECTION: HARNESSING THE RELIEF ALGORITHM AND DIVERSE AI MODELS FOR IMPROVED DIAGNOSTICS

Abstract

Early detection of breast cancer significantly improves patient outcomes and survival rates, underscoring its importance in healthcare. Early-stage breast cancer allows for less aggressive treatments and generally leads to better prognoses, as such cancers are smaller and localized, making them amenable to treatment with radiation or surgery. This study introduces a novel approach to early detection using the Relief Algorithm and advanced AI models (Multilayer Perceptron, Convolutional Neural Network, and Support Vector Machine), targeting the enhancement of diagnostic precision. Given breast cancer's role in cancer-related mortality among women worldwide, our research, utilizing the Kaggle Breast Cancer Dataset, focuses on improving early detection. We embarked on a comprehensive data preprocessing and feature selection process, employing the Relief Algorithm to identify key features for accurate classification. Our models demonstrated high accuracy in detecting breast cancer: MLP (94.68%), CNN (92.76%), and SVM (91.42%), highlighting the benefits of combining the Relief Algorithm with AI technologies in improving early detection accuracies