Interpretable machine learning model based on routine metabolic laboratory indices to identify advanced chronic kidney disease
-
By
-
March 18, 2026
Objective:
To develop and validate a machine learning model that utilizes routinely measured metabolic laboratory indices to identify advanced chronic kidney disease (CKD) in adult patients.
Key Findings:
- The Gradient Boosting classifier achieved the best discrimination with AUC values of 0.972 (internal) and 0.965 (external).
- Key predictors included urea, kidney disease type, phosphorus, albumin, and lipid-related parameters.
Interpretation:
The Gradient Boosting model effectively identifies advanced CKD using standard metabolic indices, highlighting clinically relevant metabolic patterns associated with disease severity.
Limitations:
- The study was conducted exclusively in adult patients receiving specialist care, limiting generalizability.
- The model is not intended for general population screening or longitudinal prognostic predictions.
Conclusion:
This machine learning approach could enhance risk stratification and identification of advanced CKD in clinical settings where traditional metrics may be lacking.
