Clinical Validation of AI-Powered BUN/Creatinine Ratio and Kidney Function Interpretation: A Multi-Parameter Neural Network Approach for Enhanced Renal Diagnostic Accuracy
Authors/Creators
- 1. Kantesti AI, Munich, Germany
- 2. Kantesti AI
Description
Background: Accurate interpretation of blood urea nitrogen (BUN) to creatinine ratio and associated renal biomarkers is essential for early detection of kidney dysfunction. Traditional interpretation methods rely heavily on clinician expertise and may introduce variability in diagnostic accuracy. This study validates an artificial intelligence (AI) system utilizing a 2.78 trillion parameter neural network for automated kidney function interpretation.
Methods: We conducted a retrospective analysis of 1,247,892 blood test results from 127 countries collected between January 2023 and September 2025 through the Kantesti AI platform. The AI system's interpretations were compared against gold-standard assessments by board-certified nephrologists.
Results: The AI system demonstrated 98.7% overall diagnostic accuracy (95% CI: 98.4-99.0%) for kidney function interpretation. Agreement with nephrologist assessments showed excellent concordance (Cohen's κ = 0.94).
Conclusions: AI-powered interpretation of BUN/creatinine ratio and kidney function biomarkers achieves clinical-grade accuracy comparable to specialist assessment.
Article Link: https://www.kantesti.net/bun-creatinine-ratio-kidney-function-guide/
Files
klein_et_al_2026_bun_creatinine_ai.pdf
Files
(25.3 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:7909c1850d090675312acb27f1c28b1e
|
25.3 kB | Preview Download |
Additional details
Related works
- Is supplement to
- Journal article: https://www.kantesti.net/bun-creatinine-ratio-kidney-function-guide/ (URL)
- Journal: https://www.researchgate.net/publication/399646207_JOURNAL_OF_AI_IN_CLINICAL_MEDICINE_Clinical_Validation_of_AI-Powered_BUNCreatinine_Ratio_and_Kidney_Function_Interpretation_A_Multi-Parameter_Neural_Network_Approach_for_Enhanced_Renal_Diagnostic_Accu (URL)
Dates
- Created
-
2026-01-10Publication date