Distributed Intelligence in IoHT Using Federated Learning and Edge AI for Chronic Kidney Disease Prediction Using ReLU and Bi-directional GRU with Feature-Wise Dropout and Fuzzy Neural Network

This research aims to enhance the forecasting of Chronic Kidney Disease (CKD) by discovering the importance of collective intelligence within the Internet of Healthcare Things (IoHT), by focusing on combining Edge Artificial Intelligence (AI) and Federated Learning. This method safeguards the patients’ information to stay localized, capturing the circulated flora of Federated Learning to uphold confidentiality and fulfill serious data security parameters that are challenging in Medicare backgrounds. The presence of Bi-directional Gated Recurrent Unit (Bi-GRU) with a Rectified Linear Unit (ReLU) improves the algorithm’s forecasting competencies by efficiently obtaining both the past and future needs in consecutive medical information. Moreover, the algorithm’s flexibility has been enhanced via feature-wise regressive dropout that decreases overfitting and guarantees its aptness to various datasets of patients. Thus, this agenda surpasses convolutional DL models like ALO-CNN-GRU, CNN, and LSTM exhibiting supremacy in handling complicated medical datasets with an excellent accuracy of about 99.98%. When compared to the existing algorithms, the adoption of Fuzzy Neural Network (FNN) for the forecasting of CKD efficiently reduces the computational times by allowing the algorithm to analyze irregular and confusing data. Thus, the advanced method not just allows real-time and précised findings of CKD but also proves the real-world application of Federated Learning and Edge AI for maintaining the medical data securely. The results demonstrate how this approach has the possibility to change healthcare, cheer initial detection and improve the overall results of the health by contributing extensible, efficient and patient-focused treatments inside the IoHT environment.