Toward Explainable Precision Nephrology: Machine Learning-Based Chronic Kidney Disease Prediction

Area of Science:

• * Medical Informatics
• * Machine Learning
• * Artificial Intelligence

Background:

• * Chronic kidney disease (CKD) is a progressive, incurable condition.
• * Early diagnosis significantly reduces complications and improves patient outcomes.
• * Machine learning (ML) and explainable artificial intelligence (XAI) offer potential for early CKD prediction.

Purpose of the Study:

• * To develop and evaluate ML models for reliable CKD prediction.
• * To enhance model interpretability using XAI techniques.
• * To assess the impact of feature selection and class balancing on model performance.

Main Methods:

• * Implemented various ML algorithms including ANNs, SVMs, KNN, and deep neural networks.
• * Employed feature selection methods (correlation-based, RFE, LASSO) and resampling techniques (SMOTE, SMOTETomek).
• * Evaluated models using accuracy, precision, recall, F1 Score, AUC, and Gini index, with XAI for interpretability.

Main Results:

• * KNN achieved 94.74% accuracy without SMOTE; C5.0 achieved 92.98% with SMOTE.
• * L1-regularized linear SVM showed high accuracy (89.47%) with highly correlated features.
• * Resampling improved model robustness; feature selection reduced dimensionality with minimal performance loss.

Conclusions:

• * The proposed ML framework demonstrates high accuracy and interpretability for CKD prediction.
• * Combining feature selection, class balancing, and XAI enhances model performance and clinical trustworthiness.
• * ML-based decision support systems show potential for early CKD diagnosis and personalized healthcare.