Heart failure & cardiomyopathies
Machine Learning-Based Prediction of Elevated N Terminal Pro Brain Natriuretic Peptide Among US General Population
Authors' conclusion
Using nationally representative data from the United States, both logistic regression and machine learning models well predicted elevated NT-proBNP.
Study context
- Population
- US adults aged 20–79 years without CVD, NHANES 1999–20046 prediction models
- Participants
- 10,237 adults
- Exposure
- Machine learning vs conventional regression models for predicting elevated N terminal pro brain natriuretic peptide (NT-proBNP)
- Outcome
- Prediction of elevated NT-proBNP (>125 pg/mL), AUC
Methodological detailsprediction model development and validation, nationally representative, temporal validation (trained 1999-2002, tested 2003-2004)
- Prediction model development and validation
- Nationally representative
- Temporal validation (trained 1999-2002, tested 2003-2004)
Main finding
As reported in the abstract
Elevated NT-proBNP (>125 pg/mL), AUC
| Group | AUC |
|---|---|
| SuperLearner | 0.862 (95% CI 0.847-0.878) |
| Logistic regression with splines | 0.857 (95% CI 0.841-0.874) |
AUC
SuperLearner0.862 (95% CI 0.847-0.878)
Logistic regression with splines0.857 (95% CI 0.841-0.874)
- SuperLearner vs Logistic regression: P < 0.001
- Logistic regression with splines vs Logistic regression: P = 0.08
- Prevalence of NT-proBNP >125 pg/mL
- 1510 (14.8%)
Observational data: associations, not effects.
Further findings
- Alternative NT-proBNP thresholds, SuperLearner AUC
- 0.846 (0.830-0.861) · AUC, NT-proBNP > 100 pg/mL · AUC, NT-proBNP > 150 pg/mL 0.866 (0.849-0.884)
Excluding socio-economic information
- Similar prediction performance
Original publication
Machine Learning-Based Prediction of Elevated N Terminal Pro Brain Natriuretic Peptide Among US General Population. ESC Heart Fail. Published 19 November 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →