Authors: Jayesh Sanjay Patil ,Nilesh Vani
DOI : 10.46335/IJIES.2025.10.6.14
Abstract – Chronic Kidney Disease (CKD) is a progressive, irreversible condition distinguish by a gradual decline in kidney functions, often remaining asymptomatic until advanced stages. Early detection is essential for improving patient outcomes and prolonging survival. This study shows a machine learning (ML) approach for diagnosing CKD using the CKD dataset from the UCI machine learning repository, which includes substantial missing data. To address this, K-nearest neighbors (KNN) imputation was employed, reflecting real-world clinical scenarios. Eight ML algorithms— random forest, support vector machine , logistic regression, k-nearest neighbor, AdaBoost, naive Bayes classifier, feed-forward neural network, and gradient boosting—were evaluated for their diagnostic capabilities. An additional model combining logistic regression and random forest with a perceptron was also developed, demonstrating enhanced performance across multiple simulations. The addition of AdaBoost and gradient boosting contributed to improved model robustness and predictive accuracy. These results suggest that the proposed methodology can be adapted to more complex clinical datasets, offering a valuable tool for early disease diagnosis and aiding clinicians in making timely treatment decisions.
