Machine Learning Approaches to Heart Stroke Prediction

Abstract

The ability to predict heart strokes is required in order to promote early intervention to increase chances of preventing the individuals’ death. This study aims to know which machine learning algorithms; namely Logistic Regression, Support Vector Machine (SVM), k-Nearest Neighbors (KNN) and Decision Tree classifier is more effective in terms of prediction of stroke incidence. In particular, Logistic Regression, SVM, and KNN gave impressive results which were 93% accurate. 60%, and the worst result was of Random Forest Classifier with 81. 8%. Thus, our evaluations ascertained that despite the high influence of age, hypertension and heart disease on probabilities of a stroke, conversely, those with the lowest levels of hypertension and heart disease orientation had the highest stroke likelihoods. Also, the non-smoker group had equal or higher stroke risks than the smokers with FTND scores of 4 or less; and patients with BMI between 20 and 50 also had equally higher risks of experiencing a stroke. The study also found out that other attributes such as marital status, type of residence, and type of work influenced the propensity of getting stroke. It is vital to underscore the present findings, which draw attention to the many factors that interconnect to create risk predictors for strokes and the importance of research aimed at enhancing the accuracy of these risk indicators and Identification of these risk factors would enable improved strategies and prevention measures f