- Google Research described in Nature an AI system that spots signs of high blood pressure from the pulse and motion sensors in an ordinary smartwatch, with no cuff.
- In a 196-person study it caught 65.8% of cases, against 55.3% for a first office reading. Google stresses it does not replace a diagnosis.
In short: Google Research has shown an AI system that uses an ordinary smartwatch to judge whether someone most likely has high blood pressure, with no cuff on the arm. It published the results in Nature on November 5, 2026. Google notes that more than half of adults with high blood pressure do not know they have it, because it rarely causes symptoms.
How it works
- The watch uses two sensors that are already running: an optical pulse sensor (PPG) and a motion sensor that picks up the tiny wrist vibrations caused by each heartbeat.
- The system does not try to give an exact reading at a given moment. It gathers 15-second slices of signal over several days, judges the underlying state from them, and adds the person’s height as an extra input.
- It learned from 492,347 hours of sensor data from 2,109 participants, without medical labels.
How it did in the study
The study covered 196 people at five sites in the US. Results were compared with 24-hour ambulatory blood pressure monitoring (ABPM).
- Initial screening: the watch caught 65.8% of people with high blood pressure at a specificity of 90.0%. A first office reading caught 55.3%.
- Confirmation in 48 people with a high office reading: the watch reached 78.4%, a repeat office reading 67.6%, and several days of home monitoring 89.2%.
How to read it
Google says this research helped inform the blood pressure trends feature in the Google Health app, yet it also stresses that the method is early feasibility research, not a diagnosis. The watch is meant to nudge you to see a doctor. The sample is small, and the study was run by Google with Beth Israel Deaconess Medical Center and Harvard Medical School, so I want to see independent replications. I collect more studies in the Research papers hub.