High blood pressure affects more than a billion people, and more than half of adults who have it do not know, because it rarely causes symptoms. The gap is worst among adults aged 18 to 44, who seldom get screened. Google Research has published work in Nature on a smartwatch system that flags likely hypertension from sensors the watch already carries, with no cuff and no calibration.
The approach sidesteps a problem that has defeated many wrist-based blood pressure gadgets. Instead of trying to estimate exact readings at a given moment, which shift with posture, stress and noise, it asks a simpler question: does this person have persistently high blood pressure? It combines two signals over several days. The optical pulse sensor tracks changes in blood volume at the wrist, and the accelerometer, during still periods, picks up tiny vibrations caused by the heart pushing blood into the aorta. A network pre-trained without labels on 492,347 hours of sensor data from 2,109 participants turns those signals into a profile, which is combined with the wearer's height to produce a hypertension score.
The team tested it in a prospective study at five US sites with 196 people who had no known heart disease and took no blood pressure medication, checked against 24-hour ambulatory monitoring, the clinical gold standard. As a first screen, the watch caught 65.8% of cases with 90.0% specificity, meaning few false alarms, and 80.6% of its positive flags were correct. A single office reading, by comparison, caught 55.3%. In a smaller group of 48 people with high office readings, the watch confirmed sustained hypertension with 78.4% sensitivity and 90.9% specificity, better than repeat office measurements and competitive with multi-day home monitoring.
The practical promise is a nudge. A watch that tells a 30-year-old with no symptoms to get a proper check could catch years of silent damage early. Google says the findings helped inform the blood pressure trends feature in its Google Health app.
The limits matter as much as the headline. Missing about a third of cases means a quiet watch is not a clean bill of health. The trial is small and excluded anyone already treated or with heart disease, so results may differ in the broader population. Google stresses that this is not a diagnosis and does not replace a cuff. Its blog post also calls the methods exploratory and not indicative of any product feature, which sits awkwardly with the note about the app, so check what any shipping feature actually claims before relying on it.