Integrating Cyber-Physical System with Federated-Edge Computing for Diabetes Detection and Management

Diabetes mellitus is a significant global health issue that affects millions of people. As a result, it is crucial to prioritize preventative and management strategies for this disease. In this paper, an intelligent cyber-physical system is suggested for the detection and management of type 2 diabetes. The proposed approach introduces a federated-edge computing framework, which combines the advantages of cloud computing while addressing concerns related to latency, security, and communication overhead. The architecture integrates knowledge from different Machine Learning (ML) models developed at the edge by utilizing distillation and ensemble methodologies along with the support for heterogeneous data with varied optimal aggregated models. Capitalizing on Federated Learning (FL), this process enables a Hybrid Federated Learning strategy for optimal model combination. We implementing a variety of data preprocessing methods and feature selection algorithms to pinpoint the most advantageous features for the ML models. The models used include the Support Vector Machine (SVM), Artificial Neural Network (ANN), Case-Based Reasoning with Fuzzy K-Nearest Neighbor (CBR-FKNN), and K-means combined with Logistic Regression (K-Means-LR). The efficacy of our proposed system was assessed through an experimental evaluation using the PIMA dataset.