Involving machine learning techniques in heart disease diagnosis: a performance analysis

Abstract

Artificial intelligence is a science that is growing at a tremendous speed every day and has become an essential part of many domains, including the medical domain. Therefore, countless artificial intelligence applications can be seen in the medical domain at various levels, which are employed to enhance early diagnosis and prediction and reduce the risks associated with many diseases, including heart diseases. In this article, machine learning techniques (logistic regression, random forest, artificial neural network, support vector machines, and k-nearest neighbors) are utilized to diagnose heart disease from the Cleveland Clinic dataset got from the University of California Irvine machine learning (UCL) repository and Kaggle platform then create a comparison between the performance of these techniques. In addition, some literature related to machine learning and deep learning techniques that aim to provide reasonable solutions in monitoring, detecting, diagnosing, and predicting heart disease and how these technologies assist in making health decisions are reviewed. Ten studies are selected and summarized by the authors published between 2017 and 2022 are illustrated. After executing a series of tests, it is seen that the most profitable performance in diagnosing heart disease is the support vector machines, with a diagnostic accuracy of 96%. This article has concluded that these techniques play a significant and influential role in assisting physicians and health care workers in analyzing heart patients' data, making health decisions, and saving patients' lives.