Chronic Kidney Disease Risk Estimation Using Artificial Neural Network

Abstract

Chronic kidney disease (CKD) is the most common disease of the urinary system that can threaten the survival of the human body. Early detection and lifestyle changes can prevent kidney failure and improve the chance of survival. In West Malaysia, the prevalence of chronic kidney disease is estimated to be 9% of the population. However, screening for chronic kidney disease is still neglected at the early stages. Many equations for risk estimation of kidney failure have been developed. Some of the limitations of these equations are that they may require many laboratory tests, static and not updated. In this study, a new risk estimation model for kidney disease is developed. The risk factors of kidney disease are first identified according to their energy levels, which are Low, Medium and High. The new equation is then developed based on the relationship and the estimated weight of these risk factors.Artificial Neural Network (ANN) is utilized in this study as an alternative to classic risk equations. The MATLAB software is used to train the neural network. Retrospective data from 20 subjects are used to compare the output for the conventional equation and ANN. Another 20 samples have also been generated and compared with “Kidney Disease: Improving Global Outcomes” (KDIGO) 2012 clinical guideline heat map. The results show a slight difference between the methods. The conventional method shows its capability to estimate the risk. The result also shows the potential of the artificial neural network (ANN) to improve the accuracy of chronic kidney disease risk estimation.