Comparative Analysis of Some Prominent Machine Learning Algorithm for the Prediction of Chronic Kidney Disease

Abstract

Chronic Kidney Disease (CKD) is a disorder against proper function regarding kidneys, as kidneys filter our blood whenever CKD gets worse, our blood receives wastes at a higher level, which results in sickness. It also has a substantial financial problem for families of subjects with a medical issue in  Nigeria. Among the necessary measures that need action concerning the increase of CKD is detecting the disease early and with different data mining  techniques. Data mining is gradually becoming more prevalent nowadays in healthcare, as also in fraud, abuse detection etc. Classification is a more  useful data mining function to handle items in a collection to class or target categories. For obtaining essential information from medical database,  machine learning and statistical analysis can assist in extracting hidden patterns and identify relationships from vast among of data. In this study, we  compared five (5) different models namely: Deep Neural Network (DNN), Artificial Neural Network (ANN), Naïve Bayes (NB), Logistic Regression (LR), and  K-Neighbor Nearest (KNN) to predict CKD on Gashua General Hospital (GGH) dataset. The study achieved an accuracy of 98% for DNN, KNN: 96%, NB:  97%, LR: 96% and ANN: 96%. The best performance was obtained with DNN with the highest accuracy and can be applied in real world application.