Prediction of Early Symptoms for Chronic Kidney Disease by Ensembled classifiers

Abstract

Chronic Kidney Disease is an important health issue with higher death rate.  since, there is no indications at the beginning stage of chronic kidney disease (CKD). Patients doesn’t aware of that illness. The early prediction of CKD can help the patients to recover from their infections, here the AI can implement this scope for accurate and quick process. In this paper, we purpose AI diagnoses of chronic kidney disease. The Dataset of CDK was taken from the University of California Irvine (UCI).  The missing information of patients are left due to some reasons, that missing data are filled with the KNN. After completion of incomplete qualities, six AI algorithms (Random forest, linear regression, support vector machine, Decision tree and naive Bayes classifier) were utilized to setup models. Among these AI models, support vector machine accomplished the best execution with 96.25% determination precision. By investigating the miscalculation produced by the setup models, a combined model is proposed that consolidates Random forest and naïve bayes along with utilizing perceptron, which could accomplish a normal precision of 98.75% after multiple times of reproduction. Consequently, we estimated this procedure may be material to more complex clinical information for disease finding.