An Edge-AI Based IoT Architecture for Early Disease Detection and Remote Patient Monitoring

Abstract

Edge computing and artificial intelligence are now widely used in healthcare to monitor patient data in a more rapidly and reliable way. This paper introduces a smart healthcare system that uses an Edge-AI–based Internet of Things (IoT) system for continuous health monitoring and early detection of serious diseases. The proposed system uses wearable sensors to collect data such as electrocardiogram (ECG), blood oxygen level, glucose and body temperature and it is processed on a nearby edge device instead of being sent directly to the cloud, it reduces delay and saves network bandwidth with data privacy. This paper proposes a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM) model to detect heart-related abnormalities, a combination of Random Forest and XGBoost models to evaluate the diabetes risk, and an LSTM model to monitor respiratory conditions. To run the AI models on low-power edge devices, they are optimized using model compression and lightweight deployment techniques. The proposed system uses Federated Learning to improve models without sharing raw patient data. Experimental results show that the edge-based heart disease detection model achieves an average accuracy of 97.3% while maintaining a low response time of about 145 ms. The system reduces network data transmission by about 73% compared to cloud methods. In a simulation study involving 250 synthetic patient profiles, the system successfully provided early warnings for serious health events without missing any critical cases. These results show that the proposed Edge-AI IoT system is effective and suitable for real-world healthcare monitoring.