Advancing Healthcare: Integrating A Deep Neural Network With The Bio-Inspired Puffer Fish Optimization Algorithm For Early Detection And Prediction Of Chronic Kidney Disease

Abstract

Chronic Kidney Disease (CKD) poses a significant global healthcare challenge, requiring advanced strategies for early detection and prognosis. This study introduces an innovative methodology that integrates a Deep Neural Network (DNN) with the bio-inspired Puffer Fish Optimization Algorithm (POA) to enhance CKD diagnosis and prognosis. Biomedical Sensors capture patient data, which is transmitted via the Internet of Medical Things (IoMT) for analysis. The data undergoes rigorous preprocessing, including imputation of missing values, feature encoding, data transformation, and outlier detection, ensuring dataset integrity. The processed data is used to classify CKD into various types, such as Glomerulonephritis, Hypertensive Nephropathy, Diabetic Nephropathy,Polycystic Kidney Disease, and Interstitial Nephritis, with classification optimized through POA to improve hyperparameter tuning and model performance. The DNN-POA model achieves a remarkable precision rate of 98%, offering unprecedented accuracy in CKD classification and providing insights into disease progression. This hybrid approach sets a new standard for personalized CKD management, confirming its robustness and generalizability for real-world clinical applications. However, the study is limited by its reliance on the quality and quantity of IoMT data, where variability in sensor performance and transmission could affect accuracy. Additionally, the model’s effectiveness needs validation across diverse demographic and geographic populations. This innovative hybrid strategy that incorporates deep learning techniques with POA optimization marks a significant advancement in early CKD diagnosis and personalized treatment strategies