Google announced a wrist-based feature that estimates whether a person's insulin resistance is trending higher or lower, and the company has been explicit about what the tool is not. It does not measure blood sugar, nor does it produce a glucose number.
The distinction matters because it is the one most likely to be lost. Announced at the company's August 12 product event, Insulin Resistance Trends arrives as part of a health suite called Health Guardian for eligible Pixel watches and Fitbit devices, in a market where consumers have spent years hearing that noninvasive glucose monitoring is coming to a smartwatch.
Google's own disclaimer language is unusually direct. The company states the product is "not a prescreener for diabetes" and tells users not to use it for diabetes management or to rely on it to alter any medications or insulin levels.
The Signals the Watch Actually Reads
Rather than measuring glucose, the feature analyzes physiological data the watch already collects and looks for patterns associated with how hard the body appears to be working to manage blood sugar. Google describes it as tracking monthly shifts in metabolic health using passive sensors, without requiring a blood draw.
The inputs include heart rate, heart rate variability, sleep patterns, skin temperature variation, blood oxygen, and movement. That data feeds a model that evaluates trends over a rolling window rather than producing a reading on demand, and the company says the feature alerts users when it detects sustained, elevated insulin resistance trends over the month.
The underlying research is the WEAR-ME study, published in Nature by Google researchers and collaborators. It enrolled 1,165 participants with a median age of 45 and a median body mass index of 28, and trained models against a laboratory measure of insulin resistance.
That study contains an important qualifier. Its headline accuracy, an area under the receiver operating characteristic curve of 0.80 with 76 percent sensitivity and 84 percent specificity, came from a model combining wearable data with demographics and routine blood biomarkers. A model using demographics and wearable data alone performed less well, at an area under the curve of about 0.7. The consumer feature uses wearable sensors without the blood component.
Google frames the output as a wellness trend indicator intended to prompt a user to schedule a clinical lab test if the trend worsens, not as a diagnostic result. The company's research write-up describes the method in detail.
The Gap Between the Claim and the Assumption
Insulin resistance is a real and consequential condition that can develop before blood sugar rises, which is part of why the concept appeals. The CDC reports that 115.2 million American adults have prediabetes and that 8 in 10 of them do not know it.
But a trend indicator built from heart rate and sleep data is a different kind of information from a glucose value. A continuous glucose monitor measures interstitial glucose through a sensor inserted under the skin. A laboratory test measures blood directly. This feature does neither.
The practical risk is that a reassuring trend gets read as a normal blood sugar result, or a worsening trend gets read as a diagnosis. Neither reading is supported. Google says the feature does not provide real-time clinical readings and should not be used to adjust medications or treatment plans, and that it is not intended for pregnant people.
There is also no published independent validation of the consumer implementation yet. The Nature paper supports the general approach. It does not establish how the shipped feature performs in everyday use across different bodies, skin tones, ages, and activity levels.
Who Might Reasonably Find It Useful
For someone with no diabetes diagnosis who rarely sees a clinician, a nudge toward a lab test has plausible value. Prediabetes is largely silent, and the standard screening tests, a fasting glucose or an A1C, require a visit and a blood draw that many people postpone for years.
For someone already managing diabetes, the calculus is different. Nothing about this feature substitutes for glucose monitoring, and the company says so directly. People using insulin should not use a wrist trend as information for dosing.
Anyone whose watch flags a worsening trend should treat it as a reason to book an appointment, not as a result to act on. The relevant conversation covers A1C, fasting glucose, blood pressure, weight history, and family history together.
Cost and access shape who benefits. The feature requires a compatible device; availability varies by country and device, and trends are reported monthly rather than continuously. Screening for prediabetes and diabetes is covered without cost-sharing under many insurance plans for adults who meet standard criteria, and community health centers offer testing on a sliding scale.
Pre-orders opened on the announcement date, and Google has said the Health Guardian features will arrive after launch this fall. The company's health feature documentation includes the full disclaimer language, which is worth reading before buying based on a headline.
Key Questions Answered
Does the watch measure blood sugar? No. Google states the feature is not intended to monitor glucose levels in real time and is not a prescreener for diabetes. It produces a trend, not a blood sugar number.
What does it actually measure? Physiological signals the watch already collects, including heart rate, heart rate variability, sleep patterns, skin temperature, blood oxygen, and movement, are analyzed over multiple weeks.
Can it replace a continuous glucose monitor? No. A continuous glucose monitor measures glucose levels via a sensor placed under the skin. This feature does not measure glucose at any point.
Is the approach validated? A Nature study by Google researchers supports the general method, but its strongest accuracy figures came from a model that also used blood biomarker values. The consumer feature uses only wearable sensors.
Should someone with diabetes use it for dosing? No. Google explicitly says the product should not be used to alter medications or insulin levels. Dosing decisions belong with a clinician.
What should a user do if the trend worsens? Book an appointment and ask about standard testing such as A1C or fasting glucose, rather than acting on the trend itself.
Who is it not intended for? Google says the trends should not be used by pregnant people, and availability varies by country and device.
