Google Research has shown that an artificial‑intelligence model can flag high blood pressure using only the sensors that already sit in most modern smartwatches. The system looks at the pulse‑track and motion data that the watch constantly records. It does not need a separate cuff for calibration, and it does not require a medical diagnosis to train the model. In a trial of 196 volunteers at five U.S. research sites, the AI identified 65.8 % of people with hypertension. That is close to the 55.3 % detection rate of a standard office‑based cuff measurement and similar to the performance of a cuff‑based reading taken the same day. Google says the approach has been added to the blood‑pressure trends feature in the Google Health app, but it is still labeled a research tool. The company stresses that the AI does not replace a doctor’s diagnosis; it is meant to encourage users to seek a professional check‑up.
How the model works
The model is a neural network – a type of AI that learns patterns from large data sets. Google trained it on almost 500,000 hours of smartwatch sensor recordings. The network learned to match short data segments that came from the same person, picking up individual pulse shapes and heartbeat rhythms without any labeled health information. After training, the system averages each user’s patterns over several days and outputs a probability that the person has high blood pressure. The same architecture, called SensorFM, has been used in other Google health studies that also rely on smartwatch data.
Why it matters
More than half of adults with hypertension are unaware of their condition, according to Google. A cuff‑free watch feature could alert people who would otherwise go undiagnosed, prompting them to visit a doctor sooner. The technology is still experimental, so accuracy may vary across different watch models and populations. Independent validation outside Google’s own trials has not been published yet.
Why it matters
If the watch flags a possible high‑blood‑pressure reading, a user may schedule a medical check‑up earlier than they otherwise would. However, the feature is still a research prototype and does not replace a professional diagnosis.
