An Efficient Machine Learning Approach In Diagnosing Chronic Kidney Disease

Abstract

Medical decision support systems are designed and implemented to support clinicians in diagnosis. Physicians typically work through an analysis of medical data and a knowledge base of clinical expertise. The quality of medical diagnostic decisions can be increased by improvements to these systems. The performance degradation of renal of human being is portrayed as Chronic Kidney Disease (CKD). However, the prediction of CKD is evidently done using systematic and automatic methodologies. Data mining provides a way to get the information buried in the data. They can find patterns hidden in large and complex collections of data. Ongoing collaborations between renal clinicians and computer scientists are looking at the application of data mining techniques to the area of individual patient diagnosis, based on clinical records. newlineThe existing works with machine learning algorithms fail to provide the accuracy of prediction to the needed extent. In order to fulfill the gap, this research work proposes a novel classification strategy in diagnosing CKD with additional attributes. The first methodology introduced is Optimal Fuzzy K-Nearest Neighbour (OF-KNN). The optimum performance of fuzzy is obtained by tuning the membership functions utilizing the bat optimization algorithm. The second methodology introduced is Enhanced Adaptive Neuro Fuzzy Inference System (E-ANFIS). In this study, an artificial intelligence technique is proposed for overcoming the occurrence of local minima and local maxima in diagnosing the progression of kidney disease. Normally back propagation is used in Adaptive Neuro Fuzzy Inference newlineSystem (ANFIS), but in proposed Ant Lion Optimizer (ALO) is used. newlineThe third methodology is Advanced Fuzzy Membership Support Vector Machine (AFM-SVM) to aid in clinical decision making in initiation of dialysis. The major deciding factors included are depression, work type, and alcohol intake. All the proposed techniques were executed in Matlab/Simulink platform and compared with the existing techniques. Performance