A Novel Technique for Chronic Kidney Disease Prediction using Glowworm Swarm Algorithm with Adaptive Neuro Fuzzy Inference System

The term chronic kidney disease (CKD) describes the progressive loss of kidney function; according to recent studies, the incidence of this ailment is increasing yearly. The great accuracy of machine learning approaches in diagnosing chronic kidney disease has made them more important in medical diagnosis. More recently, efforts have been made to optimize these methods by using efficient feature selection algorithms with the goal of minimizing dataset dimensionality. This research suggests using an advanced feature selection technique using Tabu Search (TS) based Stochastic Diffusion Search (SDS). 19 characteristics were chosen and 5 features were eliminated after using this method. When it comes to diagnosing CKD, the proposed Adaptive Neuro Fuzzy Inference System (ANFIS) has outperformed other state of art machine learning techniques. Through the use of an enhanced diagnostic technique utilizing the glowworm swarm optimization algorithm (GSO), this work improves the ANFIS model. By simulating glowworm behavior during food hunting, this global optimization technique increases ANFIS efficiency. Furthermore, to improve the convergence speed during network training, the suggested method incorporates a hybrid learning algorithm that combines the conjugate gradient descent and the Least Square Estimator (LSE). Fuzzy logic is added to the Adaptive Backpropagation Neural Network (ABPNN) classifier to boost performance. Findings highlight the effectiveness of the ABPNN-GSO-ANFIS in diagnosing CKD, with 99.52% accuracy, 99.34% precision, and 97.82% recall achieved.