Cross-System Meta-Analysis of Machine Learning Predictors Identifies Value-Specific Risk Drivers and Interactions Underlying Acute Kidney Injury
- 1Department of Health Outcomes and Biomedical Informatics, College of Medicine, University of Florida, Gainesville, Florida, USA
Abstract
BackgroundCurrent understanding of acute kidney injury (AKI) risk factors remains largely descriptive, offering limited precision into how specific biomarker values or physiologic thresholds influence susceptibility. We aimed to synthesize knowledge from machine learning models trained across multiple health systems to identify generalizable, value-specific risk drivers and biomarker interactions contributing to AKI risk.
MethodsWe analyzed electronic health records (EHRs) from 785,497 adult inpatients between 2010 and 2019 across nine U.S. academic medical centers within PCORnet. Interpretable gradient boosting machine models were independently developed at each health system to quantify predictor-outcome associations. Meta-regression was applied to integrate these site-level results, characterize nonlinear value-risk relationships, and identify bivariate interactions between predictors.
ResultMeta-analysis revealed consistent, value-specific risk drivers across health systems. An increase in glucose from 100 mg/dL to 140 mg/dL was associated with a 1.46-fold higher risk of AKI. Chloride and anion gap also demonstrated elevated AKI risk with risk increases overlapping portions of their reference ranges, with anion gap showing a 1.14-fold increase across 4-12 mmol/L and chloride a 1.28-fold increase across 96-100 mEq/L. Electrolytes including potassium, calcium, and sodium showed quadratic associations with AKI risk. Bivariate meta-regression identified interactions between key predictors, highlighting pathways that jointly modulate AKI risk.
ConclusionThis cross-system meta-analysis synthesizes machine learning-derived evidence into clinically interpretable knowledge, revealing how specific biomarker ranges and interactions modulate AKI risk. By moving beyond surface-level associations to quantitative, generalizable physiologic thresholds, these findings provide actionable insights to enhance risk stratification and personalized prevention in hospital care.
HighlightsO_LICross-system meta-analysis uncovered generalizable, value-specific AKI risk drivers C_LIO_LIGlucose, chloride, and anion gap within reference ranges linked to higher AKI risk C_LIO_LIKey predictor interactions suggest coordinated pathways jointly modulating AKI risk C_LI
Related
- Deep Learning-Based Spatial Profiling Identifies MET-Associated Immune Exclusion and Therapeutic Vulnerability Underlying Racial Disparities in Head and Neck Cancer
- Explainable machine learning identifies predictors for postural instability and cognitive impairment in early Parkinsons Disease
- Interpretable machine learning identifies predictors of IGF-1 during growth hormone therapy in boys
- Determine and compare limb specific predictors based on the foot pressure index of right and left foot among young adults in Saudi Arabia using machine learning models
- Sex-specific predictors of lower-limb strength: an interpretable machine learning analysis of anthropometric and body composition measures
BibTeX
@misc{chan2026crosssystem,
title = {{Cross-System Meta-Analysis of Machine Learning Predictors Identifies Value-Specific Risk Drivers and Interactions Underlying Acute Kidney Injury}},
author = {H. Y. Chan and D. Li and A. S. L. Yu and J. A. Kellum and D. Y. Fuhrman and Q. Xu and E. A. Chrischilles and L. G. Cowell and S. Chandaka and A. J. Anzalone and J. Kean and K. M. McTigue and A. S. M. Mosa and B. Taylor and M. Syed and L. R. Waitman and Y. Hu and M. Liu},
howpublished = {medRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.31.26361849},
url = {https://www.medrxiv.org/content/10.64898/2026.08.31.26361849v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Cross-System Meta-Analysis of Machine Learning Predictors Identifies Value-Specific Risk Drivers and Interactions Underlying Acute Kidney Injury
AU - H. Y. Chan
AU - D. Li
AU - A. S. L. Yu
AU - J. A. Kellum
AU - D. Y. Fuhrman
AU - Q. Xu
AU - E. A. Chrischilles
AU - L. G. Cowell
AU - S. Chandaka
AU - A. J. Anzalone
AU - J. Kean
AU - K. M. McTigue
AU - A. S. M. Mosa
AU - B. Taylor
AU - M. Syed
AU - L. R. Waitman
AU - Y. Hu
AU - M. Liu
PY - 2026
DA - 2026/08/31
PB - medRxiv
DO - 10.64898/2026.08.31.26361849
UR - https://www.medrxiv.org/content/10.64898/2026.08.31.26361849v1
AB - BackgroundCurrent understanding of acute kidney injury (AKI) risk factors remains largely descriptive, offering limited precision into how specific biomarker values or physiologic thresholds influence susceptibility. We aimed to synthesize knowledge from machine learning models trained across multiple health systems to identify generalizable, value-specific risk drivers and biomarker interactions contributing to AKI risk. MethodsWe analyzed electronic health records (EHRs) from 785,497 adult inpatients between 2010 and 2019 across nine U.S. academic medical centers within PCORnet. Interpretable gradient boosting machine models were independently developed at each health system to quantify predictor-outcome associations. Meta-regression was applied to integrate these site-level results, characterize nonlinear value-risk relationships, and identify bivariate interactions between predictors. ResultMeta-analysis revealed consistent, value-specific risk drivers across health systems. An increase in glucose from 100 mg/dL to 140 mg/dL was associated with a 1.46-fold higher risk of AKI. Chloride and anion gap also demonstrated elevated AKI risk with risk increases overlapping portions of their reference ranges, with anion gap showing a 1.14-fold increase across 4-12 mmol/L and chloride a 1.28-fold increase across 96-100 mEq/L. Electrolytes including potassium, calcium, and sodium showed quadratic associations with AKI risk. Bivariate meta-regression identified interactions between key predictors, highlighting pathways that jointly modulate AKI risk. ConclusionThis cross-system meta-analysis synthesizes machine learning-derived evidence into clinically interpretable knowledge, revealing how specific biomarker ranges and interactions modulate AKI risk. By moving beyond surface-level associations to quantitative, generalizable physiologic thresholds, these findings provide actionable insights to enhance risk stratification and personalized prevention in hospital care. HighlightsO_LICross-system meta-analysis uncovered generalizable, value-specific AKI risk drivers C_LIO_LIGlucose, chloride, and anion gap within reference ranges linked to higher AKI risk C_LIO_LIKey predictor interactions suggest coordinated pathways jointly modulating AKI risk C_LI
ER -