Paperdoi:10.1161/circ.154.suppl_1.17389
Wearable-Derived Continuous Blood Pressure Monitoring Predicts Hypertensive Crises with 72-Hour Lead Time in High-Risk Ambulatory Patients (AHA 2026 Abstract 17389)
Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 17389 (2026-11-02)
TL;DR
Abstract reports a prospective cohort of 4,190 Stage-2 hypertensive adults wearing a wrist PPG BP device (described as "BIHS grade A") for 12 months; gradient boosting on 47 hemodynamic features predicts hypertensive crisis within 72 h. Validation (20% geographically distinct hold-out): AUC 0.87, Se 82.1%, Sp 79.4%, PPV 31.6%, NPV 97.4%. Top predictors: nocturnal non-dipping (OR 4.82) and morning surge > 35 mmHg (OR 3.96).
- Form factor: wrist
Abstract
Introduction: Hypertensive crises, defined as blood pressure >180/120 mmHg with acute end-organ damage, are a leading cause of emergency department (ED) visits and cardiovascular morbidity. Office-based measurements lack temporal resolution. Cuffless wearable devices using photoplethysmography (PPG) now enable 24-hour ambulatory blood pressure (BP) monitoring, but their ability to predict acute hypertensive events prospectively remains unvalidated.
Objective: We hypothesized that machine learning applied to continuous wearable BP data could predict hypertensive crisis 72 hours in advance in ambulatory high-risk patients.
Methods: This prospective cohort study enrolled 4,190 adults with Stage 2 hypertension (aged 40-85) in continuous BP monitoring via a validated wrist-worn PPG device (BIHS grade A) for 12 months. A gradient-boosting model was trained on 80% of data using 47 hemodynamic features (nocturnal dipping, morning surge, BP variability, heart rate variability) to predict hypertensive crisis within 72 hours. External validation used a geographically distinct 20% hold-out cohort. Hypertensive crisis required ED diagnosis with end-organ damage (troponin elevation, creatinine rise >0.3 mg/dL, or acute neurological symptoms).
Results: Over 12 months, 418 crises occurred in 312 subjects (7.4%). The model achieved an AUC of 0.87 (95% CI 0.84-0.90) in the validation cohort. At the prespecified threshold (sensitivity ≥80%), sensitivity was 82.1%, specificity 79.4%, PPV 31.6%, and NPV 97.4%. The strongest predictors were nocturnal non-dipping (OR 4.82; 95% CI 3.41-6.80) and morning surge >35 mmHg (OR 3.96; 95% CI 2.88-5.44). When physicians were alerted and medications adjusted within 24 hours, crises were reduced by 61.3% and ED utilization decreased 44.7% (p=0.004).
Conclusion: Machine learning applied to wearable continuous BP monitoring predicts hypertensive crises 72 hours in advance in high-risk patients. This approach could reduce hypertensive emergencies and ED burden, warranting prospective intervention trials.
Verbatim from the publisher's DOI record (opens in a new tab) · Abstract as deposited by the publisher with Crossref; © the authors/publisher
Results on ChatPPG leaderboards
All leaderboardsNo results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.
Tasks & signals
Tasks
Signals
Keywords
How it works
Cite this paper
Jawed, I., Chaudhry, W. R., Kiyani, M., Leiva, D. R. M., Mal, M., Ikram, J., Khan, B., & Khan, Z. (2026). Wearable-Derived Continuous Blood Pressure Monitoring Predicts Hypertensive Crises with 72-Hour Lead Time in High-Risk Ambulatory Patients (AHA 2026 Abstract 17389). Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 17389. https://doi.org/10.1161/circ.154.suppl_1.17389
Jawed, Inshal, Waleed Razzaq Chaudhry, Madiha Kiyani, Daniel R Mejia Leiva, Madho Mal, Jibran Ikram, Baryalay Khan, and Zaraq Khan. 2026. "Wearable-Derived Continuous Blood Pressure Monitoring Predicts Hypertensive Crises with 72-Hour Lead Time in High-Risk Ambulatory Patients (AHA 2026 Abstract 17389)." Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 17389. https://doi.org/10.1161/circ.154.suppl_1.17389.
Jawed, I., Chaudhry, W. R., Kiyani, M., Leiva, D. R. M., Mal, M., Ikram, J., Khan, B. and Khan, Z. (2026) 'Wearable-Derived Continuous Blood Pressure Monitoring Predicts Hypertensive Crises with 72-Hour Lead Time in High-Risk Ambulatory Patients (AHA 2026 Abstract 17389)', Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 17389. doi: 10.1161/circ.154.suppl_1.17389.
Jawed, Inshal, et al. "Wearable-Derived Continuous Blood Pressure Monitoring Predicts Hypertensive Crises with 72-Hour Lead Time in High-Risk Ambulatory Patients (AHA 2026 Abstract 17389)." Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 17389, 2026. https://doi.org/10.1161/circ.154.suppl_1.17389.
I. Jawed et al., "Wearable-Derived Continuous Blood Pressure Monitoring Predicts Hypertensive Crises with 72-Hour Lead Time in High-Risk Ambulatory Patients (AHA 2026 Abstract 17389)," Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 17389, 2026, doi: 10.1161/circ.154.suppl_1.17389.
@article{jawed2026wearablederived,
title = {{Wearable-Derived Continuous Blood Pressure Monitoring Predicts Hypertensive Crises with 72-Hour Lead Time in High-Risk Ambulatory Patients (AHA 2026 Abstract 17389)}},
author = {Jawed, Inshal and Chaudhry, Waleed Razzaq and Kiyani, Madiha and Leiva, Daniel R. Mejia and Mal, Madho and Ikram, Jibran and Khan, Baryalay and Khan, Zaraq},
year = {2026},
journal = {Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 17389},
doi = {10.1161/circ.154.suppl_1.17389}
}Added to the ChatPPG research hub