Clinical Biomarker-Based Prediction of Chronic Kidney Disease Using Explainable Machine Learning

Abstract

Chronic kidney disease (CKD) is a progressive disease that needs to be diagnosed properly to slow the progression of the disease and its complications. The authors of this study suggest a clinical biomarker-based prediction framework that can be improved with explainable machine learning to enhance the accuracy and interpretability of the classification of CKD. Prior to the development of the models, the publicly available CKD dataset consisting of 400 patient records and 25 clinical attributes was preprocessed by imputing missing values, encoding categorical features, and normalizing the data. The performance of a range of supervised machine learning algorithms, namely Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbors, Naïve Bayes and Extreme Gradient Boosting (XGBoost) was assessed using the standard performance measures. The best predictive model of the models evaluated was the Random Forest classifier. Explainable Artificial Intelligence (XAI) was used to incorporate the contribution of each biomarker and hemoglobin, serum creatinine, packed cell volume, specific gravity, and albumin were found to be the most significant biomarkers. The results show how a combination of explainable ML and accessible clinical biomarkers can offer a precise, transparent, and clinically interpretable framework for early CKD diagnosis, risk stratification, and informed clinical decision making.