Analysis for Predicting Chronic Kidney Disease with the Application of Machine Learning Models

In the present scenario, chronic kidney disease is among the major diseases due to a fact that it has a high risk of translating to the end-stage renal disease with a factor of poor morbidity prognosis. It has become a global health crisis, which needs to be detected at an early stage for its prevention. The major cause of chronic kidney disease is unhealthy dietary habits & insufficient water consumption which is endangering lives. A patient's survival becomes tricky if kidneys are not diagnosed at the right time, which might lead to kidney transplantation & dialysis. It is very important to detect and diagnose this disease at a timely stage. Machine learning techniques like KNN, Decision Tree, Random forest etc. are excellent in predicting CKD. The clinical date is firstly preprocessed, then the problem of missing values is handled with an aim of feature extraction & finally applying the ML model to the test and train data. In this study, a number of variables, in addition to techniques of machine learning like decision tree, KNN & Random Forest are applied. These models also list the accuracy, F-score, precision, recall and support values. The dataset which is applied in the study is the CKD dataset, which is publicly available by the UCI platform. The highest overall accuracy obtained by the machine learning model namely, Random Forest is 97% on the test data. All the machine learning models are shown in the result section with the mentioned parameters like accuracy, f1 score, precision & recall.