Boruta Feature Selection Method for Optimizing a Case-Based Reasoning Model to Predict Heart Disease

Abstract

Nowadays, heart diseases are becoming a major problem, with which a significant part of the world population is affected. The field of medicine may significantly benefit from prediction systems using artificial intelligence techniques by making the disease prediction more accurate and faster. This paper aims to improve the predictive performance of cardiac disease diagnosis through the use of the case-based reasoning (CBR) approach, specifically focusing on its two phases: retrieval and reuse. Additionally, we aim to optimize the selection of attributes in cardiac dataset by using the Boruta method. Our approach uses various models including machine learning and deep learning models, in addition to hybrid models in the retrieval phase to accurately predict the presence or absence of a cardiac disease among patients. A robust reuse measure is used to verify the validity of the retrieved solutions and determine the necessity of applying the reuse algorithm. The results showed a significant improvement in predictive precision, with the highest accuracy achieved by the hybrid 1D CNN–SVM model on cardiac datasets. The effectiveness of the suggested approach is discussed by comparing the results with different search methods.