IOTHEALTH: EDGE-BASED TRANSFER LEARNING FRAMEWORK FOR SMART HOME HEALTHCARE

Abstract

The rising demand for continuous and personalized healthcare has led to the adoption of smart home monitoring systems that leverage Internet of Things (IoT) devices. However, processing large volumes of health data on centralized servers can result in high latency, privacy concerns, and network congestion. This study proposes IoTHealth, a deep transfer learning–based edge computing framework for efficient and real-time home health monitoring. By deploying lightweight deep learning models at the edge and leveraging transfer learning from pretrained networks, the system can accurately analyze physiological signals such as heart rate, respiration, and activity patterns while minimizing computational overhead. Experimental evaluations demonstrate that IoTHealth achieves high accuracy in health event detection, reduces data transmission to the cloud, and ensures low-latency responses for timely intervention. The framework provides a scalable, secure, and efficient solution for personalized healthcare, highlighting the potential of integrating edge computing and deep learning in smart home environments