Applying Machine Learning Techniques for Predicting the Risk of Chronic Kidney Disease

Objective: This paper aims at predicting the early detection of chronic kidney disease also known as chronic renal disease for diabetic patients with the help of machine learning methods and finally suggests a decision tree to arrive at concrete results with desirable accuracy by measuring its performance to its specification and sensitiveness. Methods: The behaviour of learning algorithms determined on a set of data mining indicators has a proportionate effect on the resulting models. Discovering the knowledge from wide databases is termed as Data mining. Besides studying the existing available Clinic Foundation Heart Disease dataset, 600 clinical records collected by us from a leading Chennai based diabetes research centre. We have tested the dataset for classification using Naïve Bayes and Decision tree method. Findings: On comparing the classification algorithms with respect to Naïve Bayes and Decision tree, we came to conclusion that the accuracy is up to 91% for Decision tree classification. Applications/Improvement: In order to increase the accuracy of the prediction result, we have utilized algorithms such as neural network and clustering data which greatly helped in our mission and also gave scope for future research.Keywords: Chronic Kidney Disease (CKD), Data Mining, Decision Tree, Diabetes, Naïve Bayes Method