Automated Stroke Prediction using Machine Learning with a Web-Based Application for Early Risk Assessment

1P LOKANADHAM, 2MERUVA THANUSRI, 3MANCHI REDDY SAHITHI,4CHITRA VISHNUVARDHAN REDDY, 5SHAIK IRFAN BASHA
1Assistant Professor, Department of Information Technology, SV college of Engineering, Tirupati, India
2B. Tech, Department of Information Technology, SV college of Engineering, Tirupati, India
3B. Tech, Department of Information Technology, SV college of Engineering, Tirupati, India
4B. Tech, Department of Information Technology, SV college of Engineering, Tirupati, India
5B. Tech, Department of Information Technology, SV college of Engineering, Tirupati, India
.Email: lokanadham.p@svce.edu.in, thanusrimeruva09@gmail.com, manchisahithi005@gmail.com,
chitravishnu8@gmail.com, irfancandy786@gmail.com
Corresponding Author/Guide: P Lokanadham, M. Tech, Assistant Professor

ABSTRACT:Stroke is a life-threatening medical condition caused by disruption of blood flow to the brain, leading to neurological damage. Early prediction and intervention are vital to reduce the severe health and economic burdens posed by stroke globally. The existing automated stroke prediction systems utilize machine learning models trained on clinical datasets, achieving reasonable accuracy in identifying high-risk patients. However, these systems face limitations including imbalanced datasets, potential data leakage, a lack of comprehensive external validation, and limited interpretability of model decisions which can hinder clinician trust and adoption. Addressing these limitations, this study proposes an advanced stroke prediction system incorporating explainable artificial intelligence (XAI) techniques such as SHAP and LIME that offer transparent and interpretable insights into the model’s decision-making process. The system integratesrobust feature selection, data balancing via SMOTE, avoidance of data leakage, and comparative evaluation of multiple classifiers including Random Forest and XG Boost, which achieve high accuracy and balanced precision-recall metrics. Furthermore, an end-to-end web and cross-platform mobile application is developed to facilitate real-time use by patients and healthcare providers, enhancing accessibility and usability for early stroke risk detection and timely intervention. KEYWORDS: neurological damage, potential data leakage, explainable artificial intelligence (XAI), SHAP and LIME,Random Forest and XG Boost, interpretability