The dream is simple: a smartwatch that measures blood glucose as effortlessly as it counts steps. For the 537 million adults worldwide living with diabetes — and the growing biohacker community tracking metabolic health — this capability would be transformative. But the physics is stubborn, and the gap between lab prototypes and wrist-ready products remains substantial.
The Technology Landscape
Current continuous glucose monitors (CGMs) like Dexcom and Abbott's FreeStyle Libre require a subcutaneous filament — a tiny needle that stays under the skin for 10–14 days. They measure glucose in interstitial fluid using an enzymatic electrochemical reaction. They are minimally invasive, not non-invasive.
True non-invasive approaches fall into several categories:
- Near-infrared (NIR) spectroscopy: Shining light through the skin and analyzing absorption spectra — glucose has characteristic absorption peaks around 1,600–2,400 nm. The challenge: water, fat, protein, and skin pigmentation all absorb in similar ranges, making signal extraction extremely noisy.
- Raman spectroscopy: Using monochromatic laser light to detect molecular vibration signatures. More specific than NIR but requires precise calibration and is sensitive to skin temperature and hydration.
- Sweat analysis: Glucose appears in sweat at roughly 1/100th of blood concentrations. Electrochemical sensors on flexible patches can detect it, but the lag time between blood and sweat glucose changes (5–15 minutes) and the influence of sweat rate complicate interpretation.
- Radiofrequency/microwave sensing: Measuring dielectric property changes in tissue as glucose concentration shifts. Promising but susceptible to motion artifacts and tissue composition variations.
The Industry Race
The stakes are enormous — the global CGM market is projected to exceed $20 billion by 2030. Key players:
- Apple: Has been working on NIR-based glucose sensing for over a decade through its Exploratory Design Group. Reports suggest the prototype is now tabletop-sized (down from room-sized), but miniaturization into a watch remains unsolved.
- Samsung: Publicly committed to non-invasive glucose in Galaxy Watch, exploring Raman-based approaches in partnership with MIT.
- Rockley Photonics: Developed a spectrophotometer-on-a-chip that measures multiple biomarkers including glucose. Their technology uses a novel "photonic integrated circuit" approach.
- Know Labs: Using radiofrequency dielectric spectroscopy (Bio-RFID™), with accuracy approaching but not yet achieving FDA requirements for glucose monitoring.
The Accuracy Gap
The FDA requires blood glucose monitors to achieve a MARD (Mean Absolute Relative Difference) below 15% — meaning the average error must be less than 15% of the reference value. Current non-invasive prototypes hover around 20–30% MARD in controlled settings, and deteriorate significantly during everyday activities (eating, exercise, temperature changes).
There is a deeper problem: even if the sensor hardware becomes perfect, physiological lag means that interstitial or sweat glucose will always trail blood glucose by minutes. For insulin dosing decisions, this delay matters.
What to Expect
The consensus among experts: we are likely 3–7 years from a consumer-grade non-invasive glucose monitor that can reliably track trends (not absolute values). The first generation will probably be positioned as a "wellness" device — showing glucose patterns and trends for non-diabetic users interested in metabolic optimization — rather than a medical device for insulin-dependent diabetics.
The bigger opportunity may not be glucose alone, but multi-analyte wrist sensors that combine glucose, lactate, hydration, and blood pressure into a metabolic dashboard — creating a new category of preventive health monitoring that goes far beyond diabetes.