Smartwatch and AI may reveal undiagnosed high blood pressure

New research indicates that a smartwatch combined with artificial intelligence may help identify high blood pressure in people who do not realize they have it.

According to the researchers, the system spotted possible hypertension with accuracy similar to an initial blood pressure screening in a clinic.

Instead of using an arm cuff to measure pressure directly, the approach relies on signals gathered by standard watch sensors. It also sidesteps the cuff calibration needed by some wearable devices, which could make it easier to use as part of everyday screening.

The goal is to encourage more people to get a proper medical evaluation before untreated high blood pressure leads to complications. Even so, the watch is meant only as a screening tool, and any result would still need to be confirmed with established blood pressure tests.

The study, published in Nature, was led by researchers including Ming-Zher Poh and Paolo Di Achille at Google Research.

It evaluated an algorithm that combines optical pulse signals, wrist movement and the wearer’s height to estimate the likelihood of hypertension.

The optical sensor measures changes in blood volume below the skin, while an accelerometer records movement. Rather than converting a single pulse reading into a standard blood pressure number, the algorithm analyzes patterns across multiple recordings.

Development used nearly 500,000 hours of smartwatch data from 2,109 people, with additional work across several groups. The main prospective validation study involved 196 adults who had no known cardiovascular disease and were not taking blood pressure medication.

The choice of comparison method is important because conventional arm cuffs can sometimes produce inaccurate readings.

Participants wore the watch during seven days of normal daily activity. Researchers then compared the smartwatch’s screening results with office blood pressure measurements, using 24-hour ambulatory monitoring as the reference standard for determining whether participants actually had hypertension.

Ambulatory monitoring uses a cuff to collect repeated readings throughout the day and while sleeping. Comparing both screening methods against that standard gave the researchers a stronger test than relying on a single clinic reading.

The smartwatch identified 65.8% of participants with hypertension, compared with 55.3% for the initial office screening. That difference was not statistically significant, so the study does not prove the watch performed better at detecting cases.

Both methods correctly classified 90% of participants who did not have hypertension. The findings also highlight why a reassuring watch result cannot rule out the condition: about one-third of affected participants were missed by the system.

The validation group was relatively small and excluded people already using hypertension medication. More research is needed to see how the approach performs in broader populations and whether alerts actually lead to diagnosis and treatment in routine care.

For now, the study shows a possible way to turn widely available wearable sensors into a prompt for medical follow-up. It points toward broader screening while leaving diagnosis, treatment decisions and follow-up blood pressure checks to clinical care.

If you are interested in this topic, please read articles about The DASH diet is a simple way to lower high blood pressure, and What should know about high blood pressure drugs and their side effects.

For more information, please see articles about Benefits of a low-sodium diet for people with high blood pressure, and Walnuts may help lower blood pressure.

Source: Google Research; University of Washington; Beth Israel Deaconess Medical Center and Harvard Medical School.