An Experimental Study and Performance Analysis of Supervised Machine Learning Algorithms for Prognosis of Chronic Kidney Disease

Abstract

In the human body, kidney clears the waste from the body and maintains vigorous balance between salt, water, and minerals in human body. The misbalancing between these leads to disturbance of normal functions of human body. Chronic kidney disease is a condition presenting the damage occurred in the normal functioning of kidneys. Early detection of chronic kidney disease helps significantly preventing severe kidney damage. The advancements in information and communication technologies certainly improves health care services for individuals and societies. In recent years, artificial intelligence and machine learning have provided potential solution for solving complex problem in variety of sectors including health care. The aim of this study is to predict the choric kidney disease from the dataset taken from the UCI repository. The dataset contains 400 instances with 25 attributes including class variable. Four state-of-the-art supervised machine learning classifiers, i.e., XGBoost, decision tree, support vector machine, and K-Neighrest Neighbor are implemented and performance is evaluated. The result shows that the XGBoost classifier outperforms with 99% value for accuracy, 100% value for precision, 97% for recall and 98% value for F1-score. The study gives a direction to develop an automated computer-assisted system for chronic kidney disease prediction and diagnosis.