Predicting Diabetes with Federated Learning: A Digital Twin and Medical Fog-based IoT Framework

In today's world, maintaining good health has become increasingly paramount. The global prevalence of diabetes has surged due to the stress of modern life and unhealthy dietary habits. Detecting diabetes at an early stage has become imperative. Leveraging advancements in Cloud and Fog computing, we can create an Internet-enabled Medical Fog framework that incorporates Machine Learning (ML) techniques to predict and diagnose diabetes at its inception. This early prediction and diagnosis would enable remote medical assistance for individuals living far from immediate medical facilities. The Internet of Medical Things allows patient data to be gathered via sensors, analyzed using ML techniques, and stored in the Cloud, providing direct access for healthcare professionals. Therefore, the current study introduces a Digital Twin (DT)-enabled Internet of Medical Fog framework, supported by Federated Learning (FL) and the SaJAYA-ANFIS ML approach for diabetes prediction. FL ensures patient data privacy is upheld while fostering a seamless, intelligent healthcare ecosystem that bridges the gap between patients and doctors through DT. Data collection begins at the Internet of Things (IoT) layer through sensors, followed by processing at the Fog layer, comprising diverse computing nodes with specific pre-processing tools and the SaJAYA-ANFIS model for diabetes prediction. Predicted outcomes are then stored in the Cloud for analysis by medical professionals. The proposed framework addresses privacy concerns through Federated Learning (FL). The proposed method has been validated with a UCI diabetes dataset and compared with state-of-the-art FL-supported ML and DT-supported ML techniques for various performance metrics (Accuracy, Precision, Specificity, Recall and Fβ-measure). The results demonstrate that our method outperforms other baselines, achieving 93.5% accuracy and 92% Fβ-measure, respectively. Fog computing, Cloud computing, Healthcare