A Survey on Machine Learning Techniques for Heart Disease Prediction

Abstract Machine learning is the field of data science that makes decisions according to the data. To process huge amounts of data, discover patterns, and find co-relations among data, this field has achieved remarkable success. It is very useful in the field of medicine, especially in cardiology. This application in the cardiology field has provided a great help to medical practitioners. Heart disease is now becoming a dangerous disease throughout the globe. Hence the main objective of this paper is to provide a review of the most frequently used ML techniques in cardiology, which ML techniques have gained maximum success in cardiology, and the overall performance of these techniques. We performed a systematic review from Jan 2018 to June 2023. We have selected 68 studies focusing on the prediction and classification of heart diseases, and heart failure too. Due to the limitations of image and signal data such as storing capacity, and noise, we have considered only text datasets for study. We have summarized why especially ML techniques are needed and their performance. The results obtained from this study show that RF, SVM, KNN, DT, LR, and NB have shown great performance in the prediction and classification of heart disease. Moreover, RF and LR have achieved great accuracy and proved more efficient than others.