AI-Enhanced ECG Model Predicts Incident Hypertension and Stratifies Risk for Adverse Outcomes

In a recent study published in JAMA Surgery, researchers developed and validated an artificial intelligence–enhanced electrocardiography (AI-ECG) risk estimator (AIRE-HTN) to predict incident hypertension and assess the risk of hypertension-related adverse outcomes. AIRE-HTN demonstrated a C index of 0.70 (95% CI, 0.69-0.71) in predicting incident hypertension in both internal and external validation cohorts, suggesting its potential to improve risk stratification beyond traditional clinical models.

Hypertension contributes significantly to global morbidity and mortality, with early identification and intervention playing a key role in reducing adverse outcomes. Although electrocardiograms (ECGs) are widely used for cardiac assessment, their role in predicting incident hypertension has been underexplored. AI-ECG models have shown promise in detecting subclinical conditions, making them a potential tool for early hypertension prediction and improved risk assessment. This study aimed to develop and validate AIRE-HTN, an AI-ECG model, to predict incident hypertension and stratify patients at risk for hypertension-related adverse outcomes.

The study was conducted as a development and external validation prognostic cohort study at Beth Israel Deaconess Medical Center (BIDMC) in Boston, MA, with external validation performed using data from the UK Biobank (UKB). AIRE-HTN was trained on 1,163,401 ECGs from 189,539 patients at BIDMC between 2014 and 2023. The algorithm was evaluated on 19,423 BIDMC patients and validated on 65,610 ECGs from 65,610 UKB participants between 2014 and 2022. AIRE-HTN used a residual convolutional neural network architecture with a discrete-time survival loss function to predict incident hypertension and stratify risk for cardiovascular events.

AIRE-HTN demonstrated consistent performance in predicting incident hypertension, with 6446 events (33%) observed in the BIDMC cohort (C index, 0.70; 95% CI, 0.69-0.71) and 1532 events (4%) in the UKB cohort (C index, 0.70; 95% CI, 0.69-0.71). The model maintained predictive performance in individuals without left ventricular hypertrophy and those with normal ECGs (C indices, 0.67-0.72). AIRE-HTN was significantly additive to existing clinical risk factors in predicting hypertension, with a continuous net reclassification index of 0.44 (95% CI, 0.33-0.53) for BIDMC and 0.32 (95% CI, 0.23-0.37) for UKB. Furthermore, AIRE-HTN independently predicted cardiovascular death (hazard ratio [HR] per standard deviation, 2.24; 95% CI, 1.67-3.00) and stratified risk for heart failure (HR, 2.60; 95% CI, 2.22-3.04), myocardial infarction (HR, 3.13; 95% CI, 2.55-3.83), ischemic stroke (HR, 1.23; 95% CI, 1.11-1.37), and chronic kidney disease (HR, 1.89; 95% CI, 1.68-2.12).

Study limitations include the use of ICD codes to define hypertension at BIDMC, which may not fully align with contemporary guidelines. The lack of validation against ambulatory blood pressure monitoring and the UKB cohort's relatively healthy population may also limit applicability to diverse populations. Although the model demonstrated only modest discriminative capability in predicting incident hypertension, its additive value to traditional risk stratification methods was notable.

“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,” the study authors concluded.

Reference

Sau A, Barker J, Pastika L, et al. Artificial intelligence-enhanced electrocardiography for prediction of incident hypertension. JAMA Cardiol. 2025;10(3):214-223. doi:10.1001/jamacardio.2024.4796