Wearable Stroke Alert System - New Health of Things Approach Based on Generative AI and Datafusion for Real-Time Stroke Monitoring

The Stroke is a severe neurological condition that significantly impacts the cerebral vascular system and is among the leading causes of hospitalization, permanent disability, and death worldwide. This study proposes an approach based on wearable devices applied to edge computing and data fusion for stroke classification through a predictive machine learning model for predicting stroke occurrences in patients, using monitoring and intelligent alerts based on generative AI for real-time pre-diagnosis generation. The new approach presents different models for stroke prediction using various data collected by wearable devices and databases. The proposed work achieved an accuracy of 99.02% and a precision of 98.38% with the Random Forest classifier combined with edge computing and data fusion, surpassing different state-of-the-art models, demonstrating satisfactory performance surpassing models in the literature with an accuracy of 99.54%, precision of 99.18%, and an AUC of 99.99% for stroke classification in patients. The model demonstrated effective performance for applications based on CAD systems applied to the health of things for different data.