Cloud-Enhanced Prediction of Stroke Outcomes Using Spiking Neural Networks

Abstract

Stroke is one of the major causes of mortality and disability worldwide, thus inducing the importance of accurately
predicting risk and aiding early intervention therapies. Conventional methods of predicting risk for stroke are
clinical evaluations, demographic factors, and the presence of any signs and symptoms; but these methods
definitely fail to assess the dynamic interactions among risk factors with each other toward stroke. In this paper,
we propose Spiking Neural Networks with cloud computing to improve both predictive accuracy and scalability
of stroke predictive models. This is further enhanced by integrating an advanced Symptom Severity Scoring and
Target Encoding approach to deliver reliable assessments of risk for stroke. Cloud technology can do this
effectively in processing and storing huge piles of datasets that could very well be updated and applied easier
across health systems. It would thus pave ways towards better-optimized, efficient, and reliable predictive tools
against strokes, which would eventually improve clinical decision-making and patient outcomes.