Edge AI-Powered Chronic Kidney Disease Prediction: Integrating Dynamic CNN-LSTM Networks and Neuro-Fuzzy Models for Advanced Diagnosis and Monitoring in IoMT-Enabled Healthcare Systems

Abstract

Background Information: Chronic kidney disease (CKD) is a worldwide health issue 
necessitating early detection for efficient management. The amalgamation of Edge AI with 
IoMT provides immediate, scalable healthcare solutions, facilitating effective disease 
prediction. The integration of deep learning with neuro-fuzzy systems improves decision
making and interpretability in resource-constrained contexts. 
Objectives: This research seeks to create an adaptive CNN-LSTM and neuro-fuzzy integration 
model for the prediction of chronic kidney disease, utilising data supplied by the Internet of 
Medical Things (IoMT). The aim is to achieve high accuracy, computational economy, and 
real-time performance while preserving interpretability for healthcare applications in Edge AI 
settings. 
Methods:The proposed hybrid model integrates CNN-LSTM for feature extraction and 
temporal pattern recognition with neuro-fuzzy logic for decision-making. It analyses IoMT 
data, enhancing computational efficiency for Edge AI. Performance was assessed by accuracy, 
precision, recall, and inference duration. 
Emphirical results: The complete model attained an accuracy of 94.6%, surpassing both 
individual and partial combinations. It exhibited a harmonious combination of precision, 
recall, and F1-score, with inference times appropriate for Edge AI healthcare applications. 
Conclusion: The hybrid model accurately predicts CKD by combining deep learning with 
neuro-fuzzy logic, so providing both precision and interpretability. Prospective advancements 
may encompass multi-disease forecasting and enhanced privacy through federated learning.