A MACHINE LEARNING METHODOLOGY FOR DIAGNOSING CHRONIC KIDNEY DISEASE AND SEVERITY ANALYSIS

Abstract

Chronic kidney disease (CKD) is a global health issue with a high rate of death and morbidity, and it is a cause of other diseases. In the early stages of chronic kidney disease, there are no evident symptoms, therefore patients frequently fail to recognise the disease. Detection of CKD at an early stage enables patients to obtain timely treatment to mitigate the disease’s progression. Due to their rapid and precise recognition capabilities, machine learning models can aid doctors in achieving this objective. In this paper, I suggest a technique for diagnosing CKD based on machine learning. The CKD data set was taken from the machine learning repository at the University of California, Irvine (UCI), which has a substantial number of missing values. Six machine learning algorithms

(logistic regression, random forest, support vector machine, k-nearest neighbour, naive Bayes clas- sifier, and multilayer perceptron) were employed to create models after effectively completing the

missing data set. The best result among these machine learning models was achieved by random forest with a diagnostic accuracy of 99 percent. By examining the errors produced by the existing ommended method is 99 percent accurate. Also, the severity of CKD is anticipated based on the individual’s clinical data. models, I presented a model that combineslogistic regression with random forest. This rec-