Polysomnography labels sleep stages with 90%+ accuracy. Consumer wearables hover at 50-70% on the same task. The gap matters. It is also smaller than it sounds for the metric most people actually care about — total sleep time. Here is what your tracker reliably measures, what it does not, and how to read the data without losing sleep over it.
How sleep is actually measured
The gold standard is polysomnography (PSG). It combines:
- EEG (electroencephalography) — brain waves
- EOG (electrooculography) — eye movement
- EMG (electromyography) — muscle tone
From these three signals, technicians (and now algorithms) classify sleep stages with high accuracy. PSG requires lab equipment. You cannot do it at home with a wristwatch.
What consumer wearables actually do
Wrist and ring devices use three less direct signals:
- Accelerometer — movement
- Photoplethysmography — heart rate and HRV via reflected light
- Skin temperature — in some devices
Machine learning models then estimate sleep stages. The models have improved markedly in the last five years. They are still working from a smaller signal set than the lab equipment.
Where accuracy lives
Validation studies consistently find consumer wearables are reasonably good at:
- Detecting that you are asleep vs awake (sensitivity is 80-95% in most studies)
- Estimating total sleep time within roughly 15-30 minutes
- Picking up major sleep disruptions
They are notably worse at:
- Distinguishing between sleep stages — REM, deep, light. Accuracy can drop to 50-70%.
- Detecting brief wake periods you do not remember
- Reliably detecting sleep disorders, including sleep apnea
Which devices are most accurate
In head-to-head validation studies, Oura, Apple Watch, and Whoop have generally outperformed Fitbit on sleep staging. Differences between leaders are small. The gap has narrowed across the category as algorithms have matured.
How to read your data
Use the trends, not the numbers. Two heuristics:
- If your tracker shows your sleep got 90 minutes worse for a week, believe it. Total sleep time and consistency are the trustworthy signals.
- If it shows you got 12 minutes of REM and yesterday you got 47, treat that with skepticism. Stage estimation is noisy on any single night.
The orthosomnia problem
Sleep clinicians have coined the term orthosomnia for anxiety driven by sleep tracker data — patients arriving convinced they have a sleep problem because their tracker said so, when their objective sleep is fine. If checking your sleep score is the first thing you do in the morning and it sets your mood, you have crossed from useful to counterproductive.
When to see a sleep specialist regardless
A wearable cannot diagnose sleep apnea. Symptoms that warrant clinical evaluation regardless of what your tracker says:
- Loud snoring
- Gasping awake
- Daytime sleepiness despite adequate time in bed
- Partner reports of breathing pauses
Bottom line
Consumer sleep tracking is genuinely useful for total sleep time, schedule consistency, and detecting big disruptions. Do not take stage breakdowns as gospel, do not let a score dictate your morning, and do not substitute the device for clinical evaluation if you have actual sleep symptoms.
