Decentralized AI For Home Health Monitoring

The growing demand for personalized and privacy preserving healthcare solutions has accelerated interest in federated learning (FL) for in-home health monitoring. This paper proposes a cloud-edge based personalized FL framework that enables intelligent data processing directly at the source—wearable and smart home devices—without transmitting raw data to centralized servers. By leveraging edge computing, the system ensures low-latency model updates while reducing bandwidth consumption and privacy risks. The proposed architecture supports personalized model training to address data heterogeneity across users,enhancing accuracy and adaptability in real-world scenarios. A case study involving elderly patients with wearable sensors demonstrates improved model performance and reduced latency compared to centralized approaches. This research highlights the potential of combining cloud-edge infrastructures with federated learning to deliver scalable, secure, and context-awarehealth monitoring systems for modern smart homes.