Wearable data predicted blood pressure crises 72 hours ahead
In a separate validation group, the model detected 82.1% of crises; 31.6% of positive predictions corresponded to a crisis.
Our read
Biomarkers- Studied:
- Cohort study of 4,190 adults with Stage 2 hypertension, 12 months (Independent replication)
- What changed:
- The model predicted hypertensive crises within 72 hours with an AUC of 0.87 (95% CI 0.84-0.90) and 82.1% sensitivity
- Why it’s aging research:
- It bears on staying healthy in midlife and later life.
Editor of record: Boris Djordjevic · AgingResearch.News
The 4,190 adults aged 40 to 85 with stage 2 hypertension, or high blood pressure, underwent continuous wrist-device monitoring for 12 months in a prospective observational cohort study. Researchers trained a machine-learning model on 80% of the data and tested it in a geographically distinct group comprising 20%. The model predicted crises within 72 hours. A crisis required an emergency department diagnosis with acute organ damage.
Across the cohort, 312 people experienced 418 crises. The strongest predictors were blood pressure failing to fall at night and a morning rise above 35 mmHg. When physicians received alerts and adjusted medications within 24 hours, this was associated with relative reductions of 61.3% in crises and 44.7% in emergency department use.
Why it matters
For people aging with high blood pressure, the findings support testing whether acting on advance warnings could reduce hypertensive emergencies and cardiovascular harm.
Caveats
Alert-driven medication changes were not randomly assigned, so differences in patient risk or care could have contributed to the reported reductions. The conference abstract did not specify the comparison behind those reductions.
- Machine learning
- Physician alerts and medication adjustment
- Wearable continuous blood pressure monitoring
- Blood pressure
- Stage 2 hypertension
- Humans
The paper
Dow University of Health Sciences; Indiana Univ SW Internal Med
AHA Scientific Sessions 2026 Abstract, Nov 2, 2026
Presented at AHA Scientific Sessions 2026, abstract 17389
Comments
Be the first to comment.