Enhanced breast Cancer Relapse Prediction Based on Ensemble Learning Approaches

Abstract

Predicting progression and deciding on the best follow-up techniques for breast cancer patients is difficult because the illness is diverse and characterized by varying relapse risks. Due to its prevalence, breast cancer has become the top cause of mortality among women worldwide, making diagnosis and prognosis particularly challenging areas of medical study. In addition, the fear of a cancer relapse is a major factor influencing cancer patients' quality of life. The study aims to help doctors determine the likelihood of a breast cancer relapse by applying ensemble learning techniques. In this research, artificial neural networks (ANN) and deep neural networks (DNN) ensembled with Weighted averaging, minority, and majority voting approaches have been investigated for performance enhancements on the breast cancer recurrence dataset sourced from the UCI-ML repository. The empirical analysis shows that this ensemble learning-enabled proposed novel approach shows improved accuracy, precision, sensitivity, specificity, and F1-score of 96.21%, 96.59%, 98.84%, 84.62%, and 97.41%, respectively. The findings of this study can aid doctors in making more informed treatment decisions, thereby improving patient outcomes.