Prediction of Heart Disease using Machine Learning Techniques

Abstract

Heart attacks and strokes account for 85 percent of these fatalities. Unhealthy food, lack of physical exercise, cigarette smoking, and excessive alcohol use are all major behavioral risk factors for CVDs. These variables can lead to high blood pressure, high blood glucose, high blood cholesterol, and obesity. It is critical to identify cardiac illness as soon as possible, as well as swiftly and correctly as possible. Complex medical data is analyzed by various data mining and machine learning techniques in literature. The findings of the in-depth examination of these research articles are extremely convincing and accurate, but the future scope of these papers reflects the need for more significant characteristics and abundant standardized data, as well as the employment of different algorithms to achieve better accuracy rates. This research paper compares Random Forest algorithm with nearest neighbor (KNN) and Naïve Bayes on standard datasets from Cleveland database and Statlog Heart Disease of University of California Irvine (UCI) repository. The major goal of the research study is to get meaningful outcomes. With only 13 characteristics, we were able to get some extremely encouraging outcomes. The results validate Random Forest Classifier with accuracy of 93.02 %, significantly outperformed Naive Bayes and KNN which have accuracy of 83.72% and 90.69% respectively.