Robotic Automation Dynamic Hybrid Neuro-Fuzzy and Deep Learning Framework with GRU-BiLSTM, Capsule Networks, Type-2 Fuzzy Logic and CNN-TCN for Accurate IoMT-Based Chronic Kidney Disease Detection

Abstract

Background Information: Chronic kidney disease (CKD) is a world health concern that requires an early detection and careful observation over time. The IoMT is offering continuously streaming data for prognostication; however, the traditional approaches fail in capturing temporal interdependencies, spatial representation of attributes, and uncertainty in data. Objectives: The objective is the engineering of a Dynamic Hybrid Neuro-Fuzzy and Deep Learning Architecture by combining GRU-BiLSTM, Capsule Networks, Type-2 Fuzzy Logic, and CNN-TCN for accurate identification of CKD, ensuring scalability, precision, and minimal latency. Methods: The proposed framework uses GRU-BiLSTM for temporal pattern analysis, Capsule Networks for spatial representation, Type-2 Fuzzy Logic for managing uncertainty, and CNN-TCN for extraction of spatial-temporal features. Data preprocessing and robust feature extraction enhance the prediction pipeline using IoMT. Results: The architecture outperforms the state-of-the-art methods, with $96.5 \%$ accuracy, $95.7 \%$ precision, and $94.9 \%$ recall, while being scalable (50 GB) and having minimal latency of 98.5 ms, making it suitable for real-time IoMT applications. Conclusion: This framework provides a strong, extensible, and adaptive solution for CKD prediction, overcoming the challenges of IoMT and improving patient care. Future work will extend this framework to other chronic diseases and secure deployments through distributed learning.