Does your watch measure deep sleep accurately?

Consumer sleep trackers are not validated against polysomnography, the reference method for assessing sleep. A recent multicentre study compared 11 consumer devices with laboratory polysomnography and found substantial variation between models in how they classified sleep stages. Some devices get deep sleep roughly right; others are wide of the mark. If the number you see in the morning does not match how you feel, the problem may lie with the device, not with you.

What the watch is actually measuring

The reference standard for identifying sleep stages is polysomnography, the laboratory method against which consumer technologies have not been validated. The American Academy of Sleep Medicine describes these consumer devices as widely used technologies that claim to measure and even improve sleep, but which lack validation against polysomnography.

That does not make the device useless. It means that what you read is an estimate, and that the quality of that estimate varies from device to device, as a multicentre study comparing several sleep trackers with laboratory polysomnography suggests.

How close to the lab does it get in practice?

A multicentre study compared 11 consumer devices — wrist-worn, bedside and phone-based — with laboratory polysomnography in 75 participants. The result was substantial variation in how they classified sleep stages, epoch by epoch. Agreement scores ranged from 0.26 to 0.69, on a scale where higher values indicate better agreement. In other words: some devices came quite close to the lab, while others were only partially consistent with it.

For deep sleep specifically, some wrist devices performed better than others. The authors concluded that some devices agreed substantially with polysomnography, while others were only partially consistent. And results varied with factors such as body mass index, sleep efficiency and the apnoea-hypopnoea index. This matters to you: the same wristband may be more reliable in one person than another, and your body is part of the equation.

There are also differences in the type of error. Wrist devices showed a high proportional bias in estimating sleep efficiency, while bedside devices showed a high proportional bias in sleep latency — the time it takes to fall asleep. None of this lets you say that your watch is accurate. It lets you say that accuracy is not a guaranteed property of the category.

Can I trust the deep sleep number I see in the morning?

You can use it as a trend indicator, with caution, but not as a clinical measurement. The American Academy of Sleep Medicine's position is clear: given the lack of validation and regulatory clearance, these consumer technologies cannot currently be used to diagnose or treat sleep disorders. They can, however, help enrich the conversation between patient and clinician, within the context of a proper assessment.

In practice, this means the following. If your watch shows a night with lots of deep sleep and you woke up feeling good, that is a pleasant coincidence, not proof. If it shows a bad night and you woke up feeling bad, that is not proof either. What you can do is look at the pattern over weeks and ask yourself whether it tracks your actual experience. If it does, it is a useful signal. If it does not, the number should not run your life.

So why does the watch sometimes seem to get it right?

An algorithm trained to recognise these patterns may come close, but the degree of agreement varies with the device and the person. The trouble is that "often" is not "always", and the error is not random: it tends to cluster in specific people and situations, as the variation with body mass index and the apnoea-hypopnoea index showed.

If you start to feel that the watch is judging your nights, it may be time to use it less.

What to do in practice with what the watch tells you

Use the device as an approximate diary, not a diagnosis. A record kept over several weeks, noting your bedtime, your wake time and how you felt, tells you more about your habits than any single deep sleep score. If you want to compare, compare yourself with yourself over time, never with someone else's numbers.

If the goal is to improve sleep, what works is not on the watch. The basics are the basics, and no device replaces them. If the watch helps you keep to a routine, good. If it becomes a source of stress, leave it on the bedside table.

What if the problem really is your sleep and not the device?

Some signs should not be managed with a watch. The NHS lists snoring or interrupted breathing during sleep — sleep apnoea — among the things that prevent a good night. It advises seeing a doctor if changing your sleep habits has not helped, if you have had difficulty sleeping for months, or if poor sleep affects your daily life. If you recognise yourself here, the next step is not to buy another device. It is to speak to a professional.

The same applies if you are pregnant, breastfeeding, taking medication or have a diagnosed condition. In those cases, a conversation with your doctor or pharmacist comes before any supplement or device.

Does a supplement come into this?

Barely, and never as a substitute for habits. What is known about consumer sleep trackers is that they vary widely from one another and are not validated for diagnosing or treating sleep disorders. If you already have a reasonable sleep routine and want to try something at night, you can talk to your pharmacist about Sleep Well Gummies. The dose is 2 gummies per day, shortly before bedtime, and they provide 1 mg of melatonin per dose. Do not exceed the recommended dose. Not recommended for children, pregnant or breastfeeding women. May cause drowsiness; avoid driving or operating machinery after taking. Contains polyols; excessive consumption may produce laxative effects. This is not a treatment for insomnia or for apnoea, and it does not replace medical assessment.

What we still do not know

We do not know whether a particular wrist device is reliable for your body in particular. We do not know whether current algorithms will be updated to come closer to polysomnography, or whether that improvement will translate into better decisions for the people who use them. And we do not know to what extent seeing the deep sleep number every morning helps or hinders people who already sleep badly. The existing evidence is enough to say that these devices do not replace a clinical examination, and not enough to say that they are useless. If you have concerns about your sleep, take the watch's record to your appointment and discuss it with your doctor. That is the use for which these data were considered suitable.