Stroke prediction using machine learning techniques

Abstract

People today are affected by a wide range of diseases as an impact of the current state of the environment and human lifestyle choices. Early detection and prediction of such diseases are necessary to prevent them from progressing to their final stages. Stroke, a cerebrovascular illness, is one of the leading causes of death and a significant financial burden on patients. Health-related behavior, which is becoming an increasingly important focus of prevention, is one of the major risk factors for stroke. The risk of stroke has been predicted using a variety of machine learning algorithms, which also include predictors such as lifestyle variables to automatically diagnose stroke. Five supervised machine learning classifiers, including Decision Tree, Random Forest, Support Vector Machine, Naïve Bayes, and K-Nearest Neighbor Algorithm are utilized in this study to predict strokes. The dataset, consisting of 5110 items with 10 attributes, is preprocessed to make it suitable for prediction, after which the aforementioned classifiers are trained on the data, and the confusion matrix is used to evaluate the performance of the classifiers. With an accuracy of 95.8%, the RF algorithm outperformed all others in the used dataset for predicting strokes based on several physiological parameters. The clinical estimation of stroke using machine learning algorithms can be more effective when compared to a person's medical background and physical activity. In addition to all of these diagnoses, stroke patients require ongoing intensive care, which can be offered by an interdisciplinary team.