A wearable will tell you, with two decimal places, how much deep sleep you had, what your recovery score is, and what your VO₂max is. The confidence of the presentation is a design choice, and it is doing a lot of work.
Some of those numbers are reliable. Some are estimates with wide error bars. And one category is close to fiction dressed as measurement.
Knowing which is which changes what you should do with the data.
What they measure well
Sleep versus wake. This is the strong result. In validation studies against polysomnography — the clinical gold standard involving EEG, eye movement and muscle activity — consumer devices detect sleep versus wake with sensitivity of 95% or above. In a Brigham and Women’s Hospital study of 35 participants, all three devices tested cleared that bar.
So total sleep time and bedtime consistency are broadly trustworthy. Those happen to be the two sleep metrics with the most evidence behind them.
Steps. Accelerometers count steps well. Given that the mortality benefit flattens between 6,000 and 8,000 steps for over-60s, approximate is entirely sufficient.
Resting heart rate, tracked over time. The absolute value may be a little off, but the trend is meaningful. A resting heart rate creeping up over weeks is a real signal — often illness, poor sleep, alcohol, or accumulated training fatigue.
What they estimate poorly
Sleep stages. This is the headline number in every sleep app, and it is the weakest.
Against polysomnography, sleep stage classification accuracy ranges from roughly 50% to 86%, depending on device and stage. In the Brigham study, deep sleep detection sensitivity was 79.5% for the best-performing device, 61.7% and 50.5% for the others. A separate study of six wrist-worn devices reported Cohen’s kappa values between 0.41 and 0.53 — moderate agreement at best.
The reason is structural. Sleep staging is defined by brain activity. Wearables have no EEG; they infer stages from movement and pulse via photoplethysmography. There is systematic underestimation of REM and overestimation of deep sleep in accelerometry-based staging.
So when your ring reports 47 minutes of deep sleep, the honest interpretation is: something in the region of that, produced by an algorithm inferring brain states from your wrist or finger. Comparing last night’s 47 minutes to the previous night’s 62 is reading noise.
HRV. Heart rate variability derived from PPG is less accurate than the ECG-derived version, particularly for high-frequency components. It is also enormously variable between individuals, which means absolute values tell you little and comparison with other people tells you nothing.
VO₂max estimates. Inferred from heart rate and pace. The absolute figure is unreliable; the direction over months, with consistent training, carries some information.
Recovery and readiness scores. These are proprietary composites, unvalidated, and different between manufacturers. They are a product feature, not a measurement.
Where accuracy gets worse — and it matters
The validation studies are conducted in healthy people without sleep disorders.
In people with obstructive sleep apnoea, insomnia or periodic limb movement disorder, device performance degrades below the published benchmarks.
That is a serious limitation, because those are exactly the people most likely to buy a sleep tracker. No consumer wearable can diagnose sleep apnoea, insomnia or any other sleep disorder. If you snore heavily, wake gasping, or feel exhausted despite adequate hours, the device is not the right tool — a GP is.
Orthosomnia
There is a recognised phenomenon in which sleep tracking makes sleep worse. Anxiety about the score, checking the app on waking, and adjusting behaviour to optimise a number that is itself an estimate.
It fits everything covered in the article on waking at 3am. Awareness of night-time wakefulness, plus interpretation of it as a problem, is precisely the mechanism that turns normal awakenings into insomnia. A device that reports your awakenings in the morning supplies that interpretation daily.
If you find yourself feeling fine but concerned because the app disagrees, the app is wrong and your body is right. That is not a joke — subjective restedness is a better guide than an inferred hypnogram.
How to use one well
Watch trends, not nights. Weekly and monthly averages are informative. Single-night numbers are noise.
Trust the simple metrics. Total sleep time, bedtime consistency, steps, resting heart rate. These are the ones that are measured rather than inferred, and they are also the ones with real evidence behind them.
Ignore the stage breakdown. You cannot control it directly, and it is the least accurate thing the device produces.
Use it as a feedback loop, not a scoreboard. The genuine value is noticing patterns: what happens to resting heart rate after alcohol, what happens to sleep duration in the weeks when work is bad. Those observations are yours, and they are more useful than any score.
Consider taking it off. If you are sleeping well and the tracker is generating anxiety, stopping is a reasonable and slightly countercultural choice.
The comparison worth keeping in mind
Set the wearable against the two previous articles in this section.
Blood pressure measured properly at home for one week costs £25 and produces a number with a diagnostic threshold behind it. An annual HbA1c costs the NHS a few pounds and tells you about the next decade.
A £300 wearable produces dozens of numbers per day, most of them estimates, some of them invented composites, none with a diagnostic threshold.
That is not an argument against wearables — the sleep-duration and step data are useful, and anything that increases activity has value. It is an argument about proportion. The device is a habit-tracking tool, and it is being sold as a diagnostic instrument.
The cheap, boring, validated measurements remain the ones that matter. That has been true of almost everything on this site.
This article is for general information and is not medical advice. Consumer wearables cannot diagnose any condition. If you have symptoms of a sleep disorder, or persistent changes in heart rate or rhythm, speak to your GP.
Sources
- Robbins R, et al. Accuracy of three commercial wearable devices for sleep tracking in healthy adults. Sensors, 2024;24(20):6532. https://www.mdpi.com/1424-8220/24/20/6532
- Performance validation of six commercial wrist-worn wearable sleep-tracking devices for sleep stage scoring compared to polysomnography. SLEEP Advances, 2025. https://academic.oup.com/sleepadvances/article/6/2/zpaf021/8090472
- Accuracy of Fitbit Charge 4, Garmin Vivosmart 4, and WHOOP versus polysomnography: systematic review. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11004611/
- Depner CM, et al., as summarised in reviews of PPG and accelerometry limitations in sleep staging and HRV derivation.
- Paluch AE, et al. Daily steps and all-cause mortality: a meta-analysis of 15 international cohorts. The Lancet Public Health, 2022.
This article is for general information and is not medical advice. If a health problem is affecting your daily life, speak to your GP.
Comments
We read every comment. Be kind, stay on topic, and please don't ask for medical advice — we can't give it.