A Comparative Analysis of Intelligent Machine Learning for Cardiovascular Disease Classification

Abstract

Machine Learning (ML) algorithms have resulted in considerable changes to health care, enabling early detection and diagnosis, including the classification and identification of Heart Disease (HD). The identification of HD through ML can aid practitioners in making accurate decisions regarding a patient’s health. This is a significant development due to HD now being the most prevalent disease worldwide, while its early diagnosis helps to save the patient’s life. ML algorithms reduce and understand HD symptoms. This study therefore proposes a novel approach differing from simple supervised ML algorithms. The research performed a comparative analysis using the dimensionality reduction algorithm known as Independent Component Analysis, as well as the ensemble technique and the Artificial Neural Network. The information employed for this analysis was obtained from the UCI ML Repository called Heart Disease. The proposed Artificial Neural Network and Adaboosting classifier demonstrated an accuracy in relation to the benchmark dataset of 0.880% and 0.821%, respectively. We thus concluded that that the dimensionality reduction of the Independent Component Analysis based classifier revealed a positive outcome, although with less accuracy than boosting and Multilayer Perceptron. To determine the performance of the algorithms, we used an accuracy score, precision, recall and F1-Score.