Chronic kidney disease prediction using boosting techniques based on clinical parameters.

Area of Science:

• Nephrology
• Computer Science
• Artificial Intelligence

Background:

• Chronic kidney disease (CKD) presents a significant global health challenge, leading to millions of deaths annually.
• Early prediction of CKD is crucial for timely diagnosis, intervention, and the development of effective preventive health strategies.
• Machine learning, particularly ensemble learning, shows promise in improving disease diagnosis and prediction accuracy.

Purpose of the Study:

• To enhance the prediction accuracy of chronic kidney disease (CKD) using ensemble learning techniques.
• To compare the performance of five distinct boosting algorithms for CKD prediction.
• To identify key features contributing to CKD development.

Main Methods:

• Utilized the boosting ensemble learning method with five algorithms: XGBoost, CatBoost, LightGBM, AdaBoost, and gradient boosting.
• Employed a CKD dataset from the UCI machine learning repository, incorporating data preprocessing, hyperparameter tuning, and feature selection.
• Evaluated model performance using accuracy, precision, recall, F1-score, Area Under the Curve-Receiver Operator Characteristic (AUC-ROC), and runtime.

Main Results:

• AdaBoost demonstrated the highest overall performance among the five boosting algorithms evaluated.
• AdaBoost achieved 100% accuracy on the training set and 98.47% accuracy on the testing set.
• The AdaBoost model also exhibited superior precision, recall, and AUC-ROC performance compared to other algorithms.

Conclusions:

• Boosting ensemble methods, particularly AdaBoost, are highly effective for accurate chronic kidney disease prediction.
• The findings suggest that AdaBoost can be a valuable tool for early CKD detection and management.
• Feature importance analysis provides insights into the key indicators for CKD, aiding in risk assessment.