Stroke Prediction using Machine Learning Methods

Abstract

Stroke, a medical emergency that occurs due to the interruption of flow of blood to a part of brain because of bleeding or blood clots. Worldwide, it is the second major reason for deaths with an annual mortality rate of 5.5 million. Every year, more than 15 million people worldwide have a stroke, and in every 4 minutes, someone dies due to stroke. A stroke is generally a consequence of a poor style of living and hence, preventable in up to 80% of the cases. Therefore, the prediction of stroke becomes necessary and should be used to prevent permanent damage by stroke. The current work predicted the stroke using the different machine learning models namely, Gaussian Naive Bayes, Logistic Regression, Decision Tree Classifier, K-Nearest Neighbours, AdaBoost Classifier, XGBoost Classifier, and Random Forest Classifier. The paper presents the comparison among all machine learning algorithms. Analysis of results revealed that the AdaBoost, XGBoost and Random Forest Classifier made the least value of incorrect predictions and had the greatest accuracy scores 95%, 96% and 97% respectively. Hence, they were the best suited model for stroke prediction and can feasibly be used by physicians to predict stroke in real world.