Google’s November 5, 2026 research account describes AI that looks for signs of persistent high blood pressure from smartwatch sensors. For wearable owners, the crucial distinction is that this is experimental screening—not evidence that a watch can diagnose hypertension on its own.
Google says the research informed Health Guardian, but cautions that its exploratory methods do not establish current or future product capabilities. The app’s Blood pressure trends feature is separately labelled for wellness, not hypertension screening or real-time readings.
An earlier author preprint was posted on December 15, 2025; November 5 is the date of Google’s account, not the original disclosure. Google announced Health Guardian’s Pixel Watch rollout on September 24, 2026.
What the study actually measured
In Google’s account, the system combines optical pulse signals and wrist-motion data across several days. It aims to classify persistent hypertension, rather than supply the two blood-pressure numbers a cuff displays. It does not need an initial cuff reading for calibration.
Google reports validation in 196 people at five US sites against 24-hour ambulatory monitoring—a portable device taking repeated blood-pressure measurements. Participants had no known cardiovascular disease and were not taking blood-pressure medication, limiting what the study establishes about people already receiving treatment.
Why 90% specificity is not 90% accuracy
The authors’ preprint abstract reports 65.8% sensitivity after seven days: the share of actual hypertension cases detected. Its 95% confidence interval was 54.0–76.3%, reflecting uncertainty in that estimate.
An arithmetic illustration makes the limitation clearer. At 65.8% sensitivity, 100 people who genuinely have hypertension would produce about 66 detections and 34 missed cases (100 − 65.8 = 34.2). These are illustrative counts, not the study’s participant totals or predictions for app users.
The reported 90% specificity describes people without hypertension: roughly 10 in every 100 would be incorrectly flagged at that rate. It does not mean “90% accurate” overall.
In a separate subgroup of 48 people who had already screened positive in a clinic, the authors reported 96.7% positive predictive value—the share of positive results that were true cases. That selected-group result is not a general accuracy rating for smartwatch owners.
Which watches have the wellness feature?
Google’s current eligibility table, checked November 5, lists:
Pixel Watch 3, 4 and 5: Blood pressure trends without a subscription.
Fitbit Edge and Fitbit Air: Google Health Premium required.
It requires Google Health app version 5.09 or later. Availability varies by country and device; a supported watch alone does not guarantee access. Check Health Guardian under the app’s Health tab for your account’s eligible features.
Google says the first monthly summary requires a 30-day wait after purchase or activation. The feature guide also requires at least five consecutive days of wear per month, with eight hours of high-quality sensor data each day.
What a result should change
Google’s guide says a “High” trend should be followed by seven days of readings with a physical cuff, shared with a physician. A “Not High” indicator does not rule out a condition. Severe symptoms such as chest pain or shortness of breath call for immediate medical care, not waiting for a monthly report.
Clinical confirmation remains separate. The US Preventive Services Task Force recommends office blood-pressure screening for adults without known hypertension, followed by out-of-office confirmation before treatment. Its guidance specifies validated, accurate upper-arm devices, with measurements taken seated after five minutes’ rest.
Do not use these trends to adjust medication or treatment; Google also says they should not be used during pregnancy.
The practical takeaway: a wearable trend can prompt a blood-pressure check; it cannot replace clinical confirmation.