Artificial Neural Network Based New Classification Methodology for Identifying Kidney Disease Risk Levels

Abstract

The healthcare sector has vast amount of medical data which are still not properly analysed; especially, discovering useful information to predict future patterns is very limited. By using data mining techniques, the current study introduced a novel classification methodology and successfully applied it in Sri Lankan domain for Chronic Kidney Disease (CKD) classifications. The current study is carried under the two phases. In the first phase, Artificial Neural Network (ANN) method namely multilayer feed-forward neural network was used to detect whether a person has a risk of having a kidney disease or not and their risk level. In the second phase, a novel forecasting methodology is proposed using multiple algorithms, which is a combination of Random Forest algorithm and an ANN hybrid methodology to detect whether a patient has fallen into a CKD or not.