Chronic Kidney Disease Detection Using Back Propagation Neural Network Classifier

Abstract

Chronic kidney disease (CKD) often referred to as chronic kidney failure considered to be life threatening disease due to huge deposition of electrolytes, fluids and waste in our body. There is a huge need to differentiate non chronic disease (NCKD) from CKD to recognize the health status of any subject visiting the nephrology clinics. This study makes use of artificial neural network (ANN) based classification of CKD and NCKD. Four attributes, Creatinine, Urea, Sodium and potassium were considered to diagnose the patient suffering from CKD or not. Datasets collected from local general hospital with n=230 was used. A feedforward back propagation neural network (BPNN) model was employed for classification and the performance of BPNN classifier was evaluated using sensitivity, specificity and classification accuracy. The Datasets was initially mined using clustering to decide the valid attribute values. The simulation results shows an overall classification accuracy 95.3% to distinguish subject suffering with CKD from NCKD.