Heart failure & cardiomyopathiesDiagnostic or prognostic study
A machine-learning model for accurate diuresis prediction in acute heart failure (DRC-AHF)
- single-centre derivation/validation study
- prospectively enrolled
- random forest regression model
- externally validated in an independent cohort
Authors' conclusion
A three-variable machine learning model (eGFR, uNa, and uCr at 2 h post-diuretic) predicted 6 h urine output with high accuracy in AHF and demonstrated adaptive improvement with expanding training data. The calculator is freely available at https://diuresis.umw.edu.pl.
Study context
- Population
- Acute heart failure (AHF) patientsprospectively enrolled AHF patients, external validation cohort n = 50 · derivation 50 · optimization 246 · validation 50
- Participants
- 296 patients
- Comparison
- Model-estimated vs observed 6 h urine output in the validation cohort
Performance
As reported in the abstract
6 h urine output, observed vs model-estimated (validation cohort) · 6 h
P = 0.480
Further findings
Classification into predefined diuresis categories (≤900, 900-1800, and ≥1800 ml)
Agreement between observed and estimated values
- correlation r
- 0.90
- 95% CI
- 0.83-0.94
- P
- P < 0.0001
- mean absolute error
- 413 ml
- mean bias (Bland-Altman)
- 67 ± 383 ml
Mean absolute error with training data size
- decrease per 100 additional patients
- approximately 200 ml
Original publication
A machine-learning model for accurate diuresis prediction in acute heart failure (DRC-AHF). ESC Heart Fail. Published 26 June 2026. doi:10.1093/eschf/xvag185
Summarised from the published abstract; every number is checked against it