Heart failure & cardiomyopathies
Machine learning to optimize use of natriuretic peptides in the diagnosis of acute heart failure
Authors' conclusion
The diagnostic performance of guideline-recommended BNP and MR-proANP thresholds for acute heart failure varied significantly across patient subgroups. A decision-support tool that combines natriuretic peptides and clinical variables was more accurate and supports more individualized diagnosis.
Study context
- Population
- Patients evaluated for acute heart failure with BNP or mid-regional pro-atrial natriuretic peptide (MR-proANP) testingFourteen studies from 12 countries
- Comparison
- Diagnostic performance of guideline-recommended BNP and MR-proANP thresholds, and of CoDE-HF, against adjudicated diagnosis of acute heart failure
- BNPn = 8,493
- MR-proANPn = 3,899
- Outcome
- Adjudicated diagnosis of acute heart failure
Methodological detailsdeveloped and externally validated a decision-support tool called CoDE-HF, machine learning
- Developed and externally validated a decision-support tool called CoDE-HF
- Machine learning
Pooled result
As reported in the abstract
Further pooled results
| Outcome | BNP | MR-proANP |
|---|---|---|
| Diagnostic performance of guideline-recommended thresholds (BNP 100 pg/mL, MR-proANP 120 pmol/L) | ||
| NPV | 93.6% (95% confidence interval 88.4-96.6%) | 95.6% (92.2-97.6%) |
| PPV | 68.8% (62.9-74.2%) | 64.8% (56.3-72.5%) |
| CoDE-HF calibration and discrimination in those without prior acute heart failure | ||
| Area under the curve | 0.914 (0.906-0.921) | 0.929 (0.919-0.939) |
| Brier score | 0.110 | 0.094 |
| CoDE-HF low- and high-probability classification | ||
| Low-probability | 30% | 48% |
| NPV | 98.5% (97.1-99.3%) | 98.5% (97.7-99.0%) |
| High-probability | 30% | 28% |
| PPV | 78.6% (70.4-85.0%) | 75.1% (70.9-78.9%) |
Diagnostic performance of guideline-recommended thresholds (BNP 100 pg/mL, MR-proANP 120 pmol/L)
- Significant heterogeneity in the performance of these thresholds was observed across important subgroups
CoDE-HF calibration and discrimination in those without prior acute heart failure
- CoDE-HF was well calibrated with excellent discrimination
CoDE-HF low- and high-probability classification
- Performed consistently across subgroups
Original publication
Machine learning to optimize use of natriuretic peptides in the diagnosis of acute heart failure. Eur Heart J Acute Cardiovasc Care. Published 12 April 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →