Your wearable measures about three things — and guesses the rest

Strap a ring to your finger or a watch to your wrist and within a night it will tell you how stressed you are, how well you recovered, and how deeply you slept. It is directly measuring almost none of that.

Underneath the dashboard, a consumer wearable reads a short list of raw signals: your pulse, from green light bounced off the blood under your skin; your movement, from an accelerometer; your skin temperature; and on some devices, a rough blood-oxygen estimate from red and infrared light. That is most of the list. Everything else on the screen — sleep stages, a stress score, a recovery number, calories, a “body battery” — is software inferring an answer from those few signals.

What the sensor actually sees

The core sensor is optical. A wearable shines light into your skin and measures how much bounces back as blood pulses through — photoplethysmography. From the rhythm of that signal it derives heart rate, and heart-rate variability, with real accuracy at rest. The accelerometer counts steps and tells when you stop moving. Skin-temperature drift is measurable. These are the honest readings, and they are genuinely useful.

Where the guessing starts

Sleep staging is the clearest example. To know whether you are in light, deep, or REM sleep, a sleep lab reads brain waves, eye movement, and muscle tone. Your wearable has none of that. It infers stages from heart rate and motion, and independent validation studies repeatedly find consumer sleep-stage accuracy falling well short of the clinical standard — fine as a rough nightly trend, unreliable as any single number. Stress scores are mostly heart-rate variability, repackaged. Calorie estimates can be off by a wide margin. “Recovery” and “body battery” are proprietary blends no one outside the company can check.

The gap matters because of how the numbers are sold. The marketing leans on phrases like “know your body” and, here and there, “medical-grade.” For the inferred metrics, that is a claim the validation literature does not support. The figure on your wrist is a model’s best guess, wearing the confidence of a measurement.

Who else is reading your body

Every one of those signals leaves your wrist. It syncs to the manufacturer’s cloud, where it is stored, analyzed, and — depending on the company and the fine print you agreed to — shared or sold. Health and life insurers have shown sustained interest in this data; employers build “wellness” programs on it. The question is no longer whether your body’s data is valuable to someone else. It is who gets it, and on what terms.

A wearable is a useful instrument if you treat it like one: watch the trends in the signals it actually measures, discount the certainty it projects onto the ones it doesn’t, and decide deliberately who else gets to see the readings. It is not a doctor. It is a sensor with a marketing department.

— Mio Flux, Health & Science