AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Why this new Fitbit feature takes us one step closer to non-invasive glucose tracking

Collected Oct 27, 2026

Google has brought two new health insights to its Fitbit Air and Fitbit Edge wearables: trends for insulin resistance and for blood pressure. The additions extend the devices from fitness and sleep tracking into territory that previously required dedicated medical hardware, and they form part of a Google suite it calls Health Guardian, described as a set of proactive safety tools spanning the Pixel Watch and Fitbit lines.

Access is gated behind Google Health Premium, the company's $100-per-year subscription. That tier includes the Google AI health coach, advanced sleep tracking and a broader set of health features, so the two new trends are effectively a perk for paying subscribers rather than something every device owner receives.

Neither feature measures insulin or glucose directly. Instead, an algorithm works from signals the wearable already collects, including heart rate, activity level and sleep, to surface changes in metabolic health. For blood pressure, the watch tracks patterns without a cuff and monitors pulse whether the wearer is asleep or awake, then delivers an insight at the end of the month. Users of the Fitbit Air or Fitbit Edge who have worn the device for at least five days qualify to see the insight in the Health tab of the app.

The insulin resistance feature follows the same pattern. It watches for shifts in metabolic health using the device's sensors, and because insulin resistance is a known precursor to type 2 diabetes, the appeal is a less invasive way to get a read on how insulin levels may be changing. The framing matters: the output points to overall trends rather than offering a diagnosis. When the feature detects patterns consistent with elevated insulin resistance, it notifies the user and prompts them to take further action.

How the model was built is the technically interesting part. Unlike a continuous glucose monitor, which penetrates the skin and reads metabolic data from blood, Google's insulin resistance feature assesses biometric and metabolic patterns through a multimodal model. According to the company's paper in Nature, that model was trained on data gathered by other wearables as well as compiled demographic data. Multimodal in this context means the model ingests several different signal types at once and learns how they combine, rather than relying on a single measurement. The trade-off is inherent: because the inputs are proxies, the output is a probability-flavored trend estimate, and its usefulness depends heavily on how well the training population matches the individual wearer.

Users who want harder numbers can connect an Abbott CGM to the Google Health app for additional glucose monitoring, so the software layer is designed to sit alongside a real sensor rather than replace one.

The pricing of the hardware is relevant to who benefits. The Fitbit Air and Fitbit Edge are positioned at $100 and $180 respectively, considerably below the flagship Pixel Watch. Bringing Health Guardian features to cheaper devices widens the pool of people who can track conditions that are common and serious, rather than reserving the capability for buyers of the most expensive hardware.

This is where the strategic reading comes in. Non-invasive glucose monitoring has been a target for years across the consumer electronics industry, with varying degrees of success and adoption, and Apple has long been rumored to be pursuing it. Google's move here is not that product. But building a multimodal model that infers biometric responses to insulin activity, and shipping it to mainstream wearables, is a step toward a monitor that relies on algorithms rather than needles and blood. In practice, each deployment also feeds the kind of real-world usage and feedback that such models need to improve.

For developers and product teams, the interesting precedent is less the health feature than the pattern: a subscription-gated, model-driven insight layered on commodity sensors, with a monthly cadence for output and a minimum wear requirement before it activates. The notification path, which nudges users toward action when risk patterns appear, also raises familiar product questions about how to communicate probabilistic health signals without implying clinical certainty. Google's own materials stop short of claiming diagnosis, and the app's framing as trends rather than readings reflects that.

Why it matters: The people most affected are Fitbit Air and Edge owners willing to pay for Health Premium, who now get metabolic and blood pressure trend signals from hardware that previously stopped at fitness tracking. For developers, it is a concrete example of turning everyday wearable signals into model-derived health insights behind a paywall, which suggests inference is becoming a product surface in its own right rather than a backend detail. It also hints, as an inference rather than a stated plan, that Google is accumulating the sensor data and model experience it would need if non-invasive glucose estimation ever becomes viable.

Read at ZDNet · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

Insulin resistance and blood pressure trends are now available on two new Fitbit devices.