Derivation and external validation of machine-learning models for risk stratification in chest pain with normal troponin
An erratum to this article was published (Eur Heart J Acute Cardiovasc Care, 28 May 2024 ↗)
Authors' conclusion
Machine-learning and LR prediction models were superior to the HEART, GRACE, and u-cTn for risk stratification of patients with chest pain and a baseline hs-cTnT <URL.
Study context
- Population
- ED patients with chest pain and initial high-sensitivity cardiac troponin T (hs-cTnT) <upper reference limit (URL)4,075 derivation
- Comparison
- Machine-learning and logistic regression (LR) models vs GRACE, HEART and u-cTn
- Outcome
- Death or MI
- Follow-up
- 90 days
Methodological detailsmachine learning
- Machine learning
Performance
As reported in the abstract
| Test or model | Value | AUC | Missed endpoint per 1000 |
|---|---|---|---|
| Gradient boosting full (GBf) | 36.7% | 0.808 | 2.2 |
| LR | 23.4% | 3.5 | |
| u-cTn | 27.2% | 3.1 |
Value
Gradient boosting full (GBf)36.7%
LR23.4%
u-cTn27.2%
AUC
Gradient boosting full (GBf)0.808
LR— not stated in the abstract
u-cTn— not stated in the abstract
Missed endpoint per 1000
Gradient boosting full (GBf)2.2
LR3.5
u-cTn3.1
- All derived models superior to HEART and GRACE scores
- P < 0.001
Further findings
Endpoint incidence
Original publication
Derivation and external validation of machine-learning models for risk stratification in chest pain with normal troponin. Eur Heart J Acute Cardiovasc Care. Published 2 August 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →