Stroke Prediction Using Machine Learning Techniques

Abstract

Most of the strokes are due to an unanticipated blocking of courses by both the brain and the heart. Detection of different stroke warning signals can help to minimize the intensity of the stroke. This research suggests an early prediction of stroke illnesses by combining the incidence of hypertension, BMI, heart disease, average glucose level, smoking status, prior stroke, and age with various machine learning algorithms. For predicting strokes, seven different classifiers were trained using these high features. Logistics Regression, Decision Tree Classifier, AdaBoost Classifier, Gaussian Classifier, K-Nearest Neighbour Classifier, Random Forest Classifier, and XGBoost Classifier were used in the research. Furthermore, the proposed study produced a 94 percent accuracy rate, with the Random Forest classifier outperforming other classifiers. This model predicts strokes with the greatest accuracy. Random Forest has the lowest false positive and false negative rates when compared to other methods. As a consequence, Random Forest is nearly the ideal classifier for predicting stroke, which physicians and patients may use to prescribe and diagnose a probable stroke early.