Synergizing GAN and Customized Neural Network for Enhanced Brain Stroke Prediction: A Web-based Implementation

Brain strokes are a global health concern, demanding effective predictive models to mitigate risks and enhance preventive interventions. Timely identification of individuals at risk is pivotal, and machine learning emerges as a promising avenue for accurate stroke prediction, potentially transforming patient care. This study introduces a comprehensive approach utilizing machine learning, addressing challenges like missing data and class imbalance through Generative Adversarial Network (GAN) for data synthesis and balancing. The custom Artificial Neural Network (ANN) tailored to brain stroke prediction achieves outstanding performance with a 99.9% accuracy. Going beyond theoretical boundaries, the research translates into practical implementation through a web-based platform for real-world stroke prediction. This platform seamlessly integrates the high-performing model, providing a user-friendly interface for individuals to assess their stroke risks. Notably, the custom ANN achieves a remarkable 95.11% accuracy even without accounting for missing data or data imbalance, underscoring its robustness in practical healthcare applications. In amalgamating advanced machine learning methodologies with practical solutions, this work significantly contributes to medical research and offers a tangible preemptive healthcare solution.