Prediction of Chronic Kidney Disease using Random Forest, XGBoost and ANN Model

Abstract

Chronic Kidney Disease (CKD) is a significant health concern worldwide, characterized by irreversible nephron loss over time. Early and accurate detection is critical. This study employs machine learning algorithms—Random Forest, XGBoost, AdaBoost, and Artificial Neural Networks (ANN)—to predict CKD. The UCI CKD dataset was utilized, with preprocessing steps like missing value imputation, scaling, and feature encoding to ensure robustness. Models were evaluated using accuracy, F1 score, precision, and recall. ANN achieved the highest accuracy (97.5%), demonstrating its capability to capture complex patterns in the data. XGBoost, while slightly less accurate, offered faster computation, making it suitable for real-time applications. The findings underscore the potential of machine learning in CKD diagnostics, paving the way for automated and accurate healthcare solutions.