Early Detection Model for Stroke Patients Through Machine Learning

Abstract

Machine learning (ML), a subfield of artificial intelligence in computer science and engineering, enables data-driven insights and predictive analysis across various domains. In healthcare, ML is increasingly applied to clinical services, therapy, and patient diagnosis, offering critical support for medical decision-making, particularly in neurological disorders. This study presents a predictive model applied to a secondary, cloud-based healthcare dataset containing clinical records. Five ML algorithms were implemented: Logistic Regression (LR), Linear Support Vector Machine (SVM), Decision Tree Classifier (DTC), Random Forest Classifier (RFC), and Gaussian Classifier (GC). The dataset was preprocessed by converting categorical string values into numerical format, and descriptive statistics and variable correlations were analyzed. Hypertension was identified as the primary risk factor for stroke. Model performance was evaluated using confusion matrix metrics, including accuracy, precision, recall, and F1-score. Among the algorithms, RFC achieved the highest accuracy of 92.6%, followed by DTC at 87%, LR at 83%, and both SVM and GC at 81%.