IoT-Enabled Health Monitoring System with Vision Transformer-Based ECG Anomaly Detection

Abstract

Early cardiovascular disease detection depends highly on real-time health monitoring because of its increasing prevalence in society. Hospital-based monitoring through periodic checkups delays disease detection causing higher death rates among patients. Existing IoT-based systems capable of tracking vital signs including heart rate and body temperature and oxygen saturation levels still need real-time anomaly alerting features and automated emergency notifications as well as AI suggestion capabilities. Deep learning models that use conventional methods like CNNs face challenges when trying to extract useful features from ECG signals which negatively affects their classification results. The proposed solution implements a Shayan Fazeli Heartbeat Dataset from Kaggle-powered IoT-based healthcare system that utilizes Vision Transformer (ViT) for heartbeat classification. The proposed model delivers 99.995% accuracy representing superior results beyond CNN-based classification methods. The system detects abnormal heart rhythms and then immediately notifies the closest hospital and generates automatic prescription messages through mobile devices for patient self-care when body temperature becomes abnormal. Our system benefits telemedicine decision-making through IoT integration with advanced deep learning which provides better emergency response time along with superior predictive accuracy and it produces improved patient outcomes.