Machine Learning in Cardiology: a Survey of Early Detection Models for Heart Diseases

Abstract

Heart disease detection and early prediction is one of the most difficult tasks in the medical field. Almost two people die every minute due to cardio vesicular diseases. According to World Health Organization (WHO), 17.9 million people depart their life every year out of which 4.77 million people are from India alone. About 13% of the world's total population is involved in cardiac disease. Early detection of the disease is crucial for effective treatment that can save millions of lives in the world. Traditional methods of heart disease detection typically involve a combination of medical history, physical examination, and diagnostic tests which are less accurate. With the advancements in machine learning and deep learning techniques, the development of accurate prediction models for heart disease has become possible. Nowadays a large volume of data is being generated in the healthcare sector, which can be leveraged to empower the development of accurate prediction models for heart diseases. Various techniques such as logistic regression, decision trees, random forest, support vector machine, artificial neural networks, and convolutional neural networks have been applied to predict heart diseases. Over the years, advancements in medical technology have led to the development of new diagnostic tools and techniques for detecting heart disease. In this study, a comparative analysis of these techniques is carried out to understand the architectures, parametric characteristics, and datasets involved in heart disease prediction. Our analysis indicates that most heart disease prediction system that have been designed using deep learning algorithms show promising performance.