Heart failure & cardiomyopathiesMeta-analysis
Machine learning to optimize the diagnostic performance of natriuretic peptides for acute heart failure across age groups
- individual patient-level data pooled across fourteen studies, random effects meta-analysis
- machine learning
Authors' conclusion
The diagnostic performance of guideline-recommended thresholds of NT-proBNP varies significantly with age. A decision-support tool incorporating NT-proBNP with age as a continuous variable provides a more consistent and accurate approach.
Study context
- Population
- Patients with suspected acute heart failurefourteen studies, median age 73 years (25th-75th percentile: 59–82)
- Participants
- 10,369 patients
- Comparison
- Diagnostic performance of guideline-recommended NT-proBNP thresholds and the Collaboration for the Diagnosis and Evaluation of Heart Failure (CoDE-HF) machine learning model across age groups
Pooled result
As reported in the abstract
Adjudicated diagnosis of acute heart failure
Patients with adjudicated diagnosis of acute heart failure
43.9% (4549/10 369)
Further pooled results
- CoDE-HF diagnostic performance across all age groups
- 96.4% to 99.5% (93.8-99.8% CIs) · NPV · PPV 81.1% to 84.2% (74.7-90.4% CIs)
- NPV of the rule-out threshold of 300 pg/mL, by age: patients ≥80 years NPV 88.7% (CI 84.2–92.1%)
- NPV of the rule-out threshold of 300 pg/mL, by age: <50 years NPV 98.9% (CI 97.6–99.5%)
- PPV of age-stratified rule-in thresholds, by age: <50 years PPV 62.0% (CI 56.2–67.5%)
- PPV of age-stratified rule-in thresholds, by age: ≥80 years PPV 79.6% (CI 70.7–86.3%)
CoDE-HF diagnostic performance across all age groups
- CoDE-HF was more accurate than guideline-recommended thresholds across all age groups
Original publication
Machine learning to optimize the diagnostic performance of natriuretic peptides for acute heart failure across age groups. ESC Heart Fail. Published 8 January 2026. doi:10.1093/eschf/xvaf006
Summarised from the published abstract; every number is checked against it