Comparative Analysis of Machine Learning Algorithms for Enhancing Stroke Prediction in Healthcare Database

Stroke is a major global cause of disability and death, often resulting from sudden blockages in the brain or heart. Recognizing early warning signs can help reduce stroke risk and is crucial for effective stroke prevention and management. This study enhances stroke prediction using machine learning and neural networks, analyzing risk factors like hypertension, heart disease, ever-married, work type, residence type, glucose levels, BMI, smoking, gender, and age. Ten machine learning algorithms are tested on a well-curated clinical dataset. The results showed high accuracy for Random Forest (98.17%), Logistic Regression (98.32%), CatBoost (98.52%), and an Ensemble Technique (98.63%). However, the Artificial Neural Network (ANN) outperformed all with 99.37% accuracy, the highest AUC score, and the lowest false positive rate. These findings highlight ANN’s effectiveness in stroke risk assessment, offering valuable insights for improved prevention and timely medical interventions.