Feature Selection-based Machine Learning Comparative Analysis for Predicting Breast Cancer

Abstract

Breast cancer is a serious disease, and therefore early detection is crucial for successful treatment and patient management. Unfortunately, globally, the number of breast cancer cases is increasing due to various multifaceted factors. It is currently one of the leading causes of cancer deaths in women, worldwide. Cancerous cells in the breast can form lumps that impact the patient’s health, and even seemingly harmless tumors could be fatal if undiagnosed early enough. Fortunately, artificial intelligence techniques have proven effective in detecting diseases, and doctors can therefore use them to effectively and accurately diagnose breast cancer early. This paper explores the use of genetic algorithms, ant colony optimization, and Hybrid Hopfield Neural Network-E2SAT (HHNN-E2SAT) models, for breast cancer prediction. The HHNN-E2SAT models outperform standard algorithms like the Random Forest and Support Vector Machines, achieving over 98% on all performance metrics (i.e. Accuracy, F1-score, Sensitivity, Specificity, and Precision).