Artificial Intelligence–Enhanced Electrocardiography for Prediction of Incident Hypertension
Authors' conclusion
Results suggest that AIRE-HTN, an AI-ECG model, can predict incident hypertension and identify patients at risk of hypertension-related adverse events, beyond conventional clinical risk factors.
Study context
- Population
- Patients with routinely collected ECGs at Beth Israel Deaconess Medical Center (BIDMC) (Boston) and UK Biobank volunteers
- Participants
- 189,539 BIDMC training patients
- Comparison
- BIDMC vs UK Biobank (UKB)
- BIDMCn = 19,423
- UKBn = 35,806
- Outcome
- Incident hypertension (predicted by AIRE-HTN)
Methodological detailsdevelopment and external validation, external validation (UK Biobank), secondary care setting
- Development and external validation
- External validation (UK Biobank)
- Secondary care setting
Performance
As reported in the abstract
Incident hypertension
| Test or model | Value | C index |
|---|---|---|
| BIDMC | 33% | 0.70 (95% CI, 0.69-0.71) |
| UKB | 4% | 0.70 (95% CI, 0.69-0.71) |
Value
BIDMC33%
UKB4%
C index
BIDMC0.70 (95% CI, 0.69-0.71)
UKB0.70 (95% CI, 0.69-0.71)
Further findings
Hypertension-associated adverse outcomes (adjusted Cox models)
| Estimate | ||
|---|---|---|
| AIRE-HTN scoreCV death · beyond traditional risk factors | HR per standard deviation 2.24 (95% CI 1.67–3.00) | |
| AIRE-HTN scoreHF · beyond traditional risk factors | HR 2.60 (95% CI 2.22–3.04) | |
| AIRE-HTN scoreMI · beyond traditional risk factors | HR 3.13 (95% CI 2.55–3.83) | |
| AIRE-HTN scoreIschemic stroke · beyond traditional risk factors | HR 1.23 (95% CI 1.11–1.37) | |
| AIRE-HTN scoreCKD · beyond traditional risk factors | HR 1.89 (95% CI 1.68–2.12) | |
| 0.20.5125Ratio (log scale) |
| Outcome | Continuous net reclassification index (95% CI) |
|---|---|
| Incremental value over existing clinical risk factors for incident hypertensionAIRE-HTN vs existing clinical risk factors · BIDMC | 0.44 (0.33–0.53) |
| Incremental value over existing clinical risk factors for incident hypertensionAIRE-HTN vs existing clinical risk factors · UKB | 0.32 (0.23–0.37) |
Incident hypertension in individuals without LV hypertrophy and with normal ECGs (subgroups)
- Performance maintained (C indices)
- 0.67-0.72
Original publication
Artificial Intelligence–Enhanced Electrocardiography for Prediction of Incident Hypertension. JAMA Cardiol. Published 1 March 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →