Date of Publication :28th June 2024

Abstract:Using a dataset that includes lifestyle indicators, medical history, and demographic information, this study attempts to predict the risk of stroke with machine learning algorithms. Feature engineering and management of categorical variables are part of data preprocessing. Various techniques for classification are trained, such as Random Forest Classifier, Decision Tree Classifier, K-Nearest Neighbors (KNN), and Logistic Regression. Classification reports and accuracy ratings on training and testing datasets are used to evaluate the performance of the model. Furthermore, the Random Forest Classifier's feature importance analysis pinpoints important risk variables. The Random Forest Classifier outperforms the others, according to the promising accuracy results. Based on easily accessible demographic and health-related data, this shows that machine learning techniques can support early stroke risk assessment.

Reference :