Chronic Kidney Disease Detection Using Machine Learning Techniques

Abstract

The aim of this study was to test the use of machine learning in early detection of Chronic Kidney Disease (CKD). The main problem was that CKD usually progresses slowly without any symptoms and in many cases the patient does not realize it in the early stages. Therefore, timely and accurate screening requires machine learning models that can differentiate between healthy and unhealthy patients. In this thesis, we used four machine learning classifiers - logistic regression, decision tree, random forest, and KNN. Each model was trained with the same dataset and preprocessing, which solved missing values and other problems. The tests showed that random forest performed the best, giving completely correct results (in accuracy, precision, recall, and F1-score). However, other models also gave good results, of which logistic regression and decision tree are particularly noteworthy. These results show that machine learning can improve the accuracy of CKD detection and can help improve the clinical screening process. In the future, studies with larger and more diverse datasets can be conducted to further improve the performance of the model.