Google Patents Heart-Rate Sensor Adjustments Based on Your Skin Tone
Heart-rate sensors in smartwatches have long struggled to read accurately across different skin tones. Google is filing a patent that would let a Pixel Watch adjust its light sensors before it even takes a reading.
How Google's skin tone fix improves wearable health accuracy
Most smartwatch health sensors work by shining light into your skin and measuring how much bounces back. The problem is that skin tone affects how light is absorbed and reflected, and most devices use a one-size-fits-all setting that works better for some users than others.
Google's approach here is to have your smartwatch pull skin tone information from photos already stored on your phone, then use that information to tune the sensor before it measures your heart rate or blood oxygen. The watch would adjust things like the brightness or color of the light it emits, so the reading is better matched to your skin.
You wouldn't have to do anything manually. The system is designed to happen in the background, using photos you've already taken, without requiring a separate skin tone test or setup step.
… obtaining, from a mobile device connected to the wearable device, skin tone data indicating a skin tone of the user; and adjusting a parameter of the PPG sensor based on the skin tone data.
Translation: The watch gets your skin color from your phone and changes how the heart monitor works.
How the Pixel Watch would pull and apply skin tone data
The patent describes a photoplethysmography (PPG) sensor calibration system. PPG is the technology inside nearly every fitness wearable: it shines LED light (usually green, red, or infrared) against your skin and detects how the reflected light changes with each heartbeat. The amount of light your skin absorbs depends partly on melanin concentration, which varies with skin tone, and that variation can throw off readings if the sensor isn't tuned appropriately.
Here's the core flow the patent describes:
- The wearable receives a signal that a health measurement is about to begin.
- It queries the paired mobile device (your phone) and retrieves skin tone data derived from photos stored there.
- It adjusts a parameter of the PPG sensor, specifically things like LED power output (how bright the light is) or wavelength selection (which color of light to use), based on that skin tone data.
The skin tone analysis happens on the companion device, not the watch, which makes sense given that phones have more processing power and already store photos. The watch receives a derived skin tone value and acts on it locally.
The patent doesn't specify exactly how the photo analysis works, only that skin tone information is derived from photos and passed to the wearable as an input to the calibration step.
Techniques of calibrating a PPG sensor include accessing photos on a companion device connected to a wearable device having PPG sensors and deriving skin tone information from the photos.
Translation: The system looks at pictures on your connected device to figure out your skin tone for better accuracy.
What this means for health equity in wearable devices
Accuracy gaps in PPG sensors across skin tones are a documented problem, and they've drawn regulatory and academic attention in recent years. A sensor that is miscalibrated for a particular user could return heart rate or blood oxygen readings that are subtly off, which matters more as wearables edge toward medical-grade health monitoring.
Google's ongoing filings around wearable health sensing suggest the company sees health accuracy as a product differentiator. For everyday users, the promise is straightforward: your watch would do more to account for your body before it takes a reading, rather than treating everyone as a default baseline.
This is the 11th Google filing we've tracked since May in our wearables reading your body watchlist, following gesture reading via cameras and earbuds tracking chewing and sleep.
The system estimates your skin tone from photos on your phone rather than measuring it directly at your wrist, and that gap is where things can go wrong. A selfie taken under bad lighting or at an odd angle could feed the watch a flawed estimate, and the watch has no way to catch that mistake.
That is a real cost, and the trade is defensible only if close-enough estimates still improve accuracy often enough to matter. Asking users to hold still for a dedicated skin measurement during setup would almost certainly produce better inputs, but most people would skip it.
The quieter risk is that a watch using a bad estimate gives users confidence in readings that are no more accurate than before. A system that sometimes works well and sometimes fails silently is harder to trust than one that just admits its limits.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings
5 drawing sheets from US 2026/0262975 A1 · click any drawing to enlarge
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →