Bayesian Optimized Xgboost for Enhanced Chronic Kidney Disease Prediction and Feature Importance Analysis
Year: 2025
Type: article
Abstract: Chronic Kidney Disease (CKD) poses an intricate challenge to global healthcare. Timely prediction is critical for effective patient management on an individual level. Existing approaches to predicting CKD tend to apply more traditional techniques with systematic and linear methodologies, resulting in lower accuracy as they miss complex, non-linear relationships within the individual patient data. In this study, we offer a new approach that evaluates CKD risk using eXtreme Gradient Boosting (XGBoost) machine learning and improves it further with Bayesian Optimization. The method comprises of collecting and merging clinical, demographic, and laboratory data for 400 patients. Accomplished with the proposed model is thorough preprocessing, applying generalization through cross-validation, stepwise data partitioning into training and test subgroups, and avoidance of data leakage. For these reasons, the Bayesian Optimized XGBoost Model achieved an astonishing 98.8% prediction accuracy on the test set.
Institutions: Landmark University, Federal University of Technology Minna
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W7125579669