AI Tool Predicts Hypertensive Crisis Up to 72 Hours Ahead

It can lead to stroke, heart attack, kidney injury, aortic dissection, or pulmonary edema, making early detection critical. Because many patients experience no warning symptoms before serious complications develop, identifying risk in advance remains difficult. A new study now shows that AI combined with wearable monitoring may detect elevated risk up to three days before a hypertensive crisis occurs.

Researchers at the University of Maryland Capital Region Medical Center (Largo, MD, USA) evaluated an AI tool that analyzed data from a portable cuff-based blood pressure monitor and a wrist-worn fitness device. The blood pressure monitor collected repeated readings during daily activities and sleep, while the wrist-worn device used light-based sensors to track cardiovascular signals including heart rate and heart rate variability.

The AI model analyzed 47 characteristics from the monitoring data, including nighttime blood pressure dipping, morning blood pressure surges, changes over time, and heart rate variation. When the model identified an increased risk of hypertensive crisis within 72 hours, it generated an alert for the physician. The alert prompted clinical follow-up, symptom review, and medication adjustment when needed.

The 12-month study included 4,190 adults ages 40 to 85 with stage 2 hypertension, defined as blood pressure of at least 140/90 mm Hg. Participants had an average age of 62.4 years, and 46% were women. All received care at a single hospital in Sindh, Pakistan, and were followed from January 2025 to January 2026.

During the study, 312 participants experienced 418 hypertensive crises, representing about 7% of the cohort. The tool distinguished patients at higher versus lower risk up to 72 hours in advance. About 32% of participants with a positive alert experienced a hypertensive crisis, while 97.4% of those without an alert remained crisis-free. Absence of a normal sleep-related blood pressure dip was associated with nearly five times the odds of an upcoming crisis, while a morning blood pressure rise greater than 35 mm Hg was linked to nearly four times the odds.

When physicians received an alert and adjusted medication within 24 hours, hypertensive crises were about 61% lower and emergency department visits about 45% lower than among patients without timely alert-triggered intervention. The findings are scheduled for presentation at the American Heart Association’s Scientific Sessions 2026, taking place November 6–9 in Chicago. The researchers noted that the single-hospital study in Pakistan will require validation in more diverse populations and care settings before broader use.

Related Links
University of Maryland Capital Region Medical Center 
American Heart Association