Edge Split Smart Health 5G in Real Time AI Driven Patient Monitoring

Abstract

Real-time monitoring of patient health is an essential feature of modern smart healthcare that cannot be implemented at once. Conventional cloud-based solutions usually have problem with response time, power consumption, and confidentiality when managing large volumes of physiologic information delivered by wearable computers. Our proposal in this work is Edge-Split Smart Health 5G (ESS-5G), a new architecture combining wearable IoT sensors, 5G network slicing, multi-access edge computing (MEC) and a hybrid AI engine that comprises federated learning, transfer learning, and adaptive model scheduling. ESS-5G can record vital signs, physiological measurements, and motion patterns on a finegrained basis and dynamically offload the computations on edge and cloud levels depending on real-time network conditions of 5G, energy budgets of devices, and patient contexts. The framework introduces lightweight encryption and secure network-sliced channels to achieve the privacy and safety of data. We tested ESS 5G in 500 patients with ECG, SpO 2, and accelerometers. End-to-end latency of 10ms or less is shown, detrimental object detection of 97.3 per cent, and resource savings up to 30 per cent versus fundamental 4G/AI systems. The proposed system can help solve heterogeneous device interoperability challenges, control energy consumption by AI deployment, and privacy-preserving analytics, which offers a scalable and secure solution that meets Industry 4.0 paradigms in next-generation intelligent healthcare.