Heart Disease Prediction using Machine Learning and Deep Learning Approaches: A Systematic Survey

Abstract

Cardiovascular disease is still a major killer around the world and is caused by inappropriate life styles and nutritional processes. Early diagnosis and subsequent prevention of diseases have ensured that healthcare is a significant research area, more so where heart diseases are concerned. Due to the expansion of electronic data in today’s world, healthcare departments are overburdened with extensive varieties of medical data, which requires proper modes of managing and extracting valuable information. This article performs a systematic literature review of the data mining, machine learning and deep learning approaches used for heart disease prediction, especially in proving the efficiency of the techniques in identifying the hidden patterns in the large health datasets for early diagnosis. It shows how these computational methods help to find patterns related to risks and prevent them with the needed interventions. Furthermore, common issues with the existing prediction models are discussed, including the lack of focus on interpretability, generalizabity, and scalability of the models. This survey concludes by presenting directions for future research to reduce the error rate, real-time analysis in a clinical environment, and personalized analysis. This guidance will be useful to plan better and improve these directions to allow the reduction of deaths due to this pathology.