Abstract: This study explores the use of machine learning to improve early detection of Chronic Kidney Disease (CKD). Using a data set from the UCI repository, seven classifiers and multiple feature selection techniques were evaluated. The Linear SVM with L2 regularization achieved 98.86% accuracy with SMOTE and full features, while a Deep Neural Network reached the highest accuracy of 99.6%. The results highlight the effectiveness of machine learning, especially deep learning, in enhancing CKD diagnosis.

Keywords: predictive modeling, SVM, logistic regression, neural network, random tree.

Downloads: | DOI: 10.17148/IARJSET.2025.12505