PREDICTING CHRONIC KIDNEY DISEASE USING MACHINE LEARNING ALGORITHMS

Abstract

In today's world, while many people are conscious of their health, their busy schedules and demanding
workloads often lead them to only address health issues when symptoms become apparent. Chronic Kidney Disease
(CKD) poses a particular challenge because it often has no obvious symptoms, making it hard to predict, detect, or
prevent, and potentially leading to significant long-term health problems. However, machine learning (ML) offers a
promising solution due to its strengths in prediction and analysis. This paper explores nine different ML techniques,
including K-nearest neighbors (KNN), support vector machines (SVM), logistic regression (LR), Naïve Bayes, Extra
Trees Classifier, AdaBoost, XGBoost, and LightGBM. These models were developed using a dataset from
Kaggle.com that contains 14 features and 400 records related to chronic kidney disease. The study evaluates the
performance of these models and demonstrates that the LightGBM model achieves an unprecedented accuracy rate
of 99.00% in predicting CKD.