Paper Title
HEART STROKE PREDICTION USING NEURAL NETWORKS: A COMPREHENSIVE APPROACH

Abstract
Stroke remains a significant global health challenge, underscoring the urgent need for robust predictive models to support early intervention and improve patient outcomes. In this study, we introduce a custom convolutional neural network (CNN) specifically engineered for stroke risk prediction, leveraging a rich healthcare dataset. Our research addresses critical data challenges, implementing a tailored preprocessing pipeline that effectively manages missing values, transforms categorical variables, and applies SMOTE to balance classes—an essential step in ensuring accurate predictions. The proposed CNN model, featuring dense layers and dropout regularization, achieved a notable 91% accuracy, outperforming traditional machine learning techniques and demonstrating the capacity of deep learning to revolutionize healthcare predictions. These findings provide a solid foundation for advancing predictive models in healthcare, offering an adaptable framework that can inspire future research to enhance accuracy and support timely medical interventions. Keywords - Stroke Prediction, CNN, Healthcare data, Prediction model, SMOTE, Deep Learning