On a typical night, I go to bed wearing a tracking watch on one wrist and another type of tracker on the other wrist. Then I compare their sleep findings while I drink my morning coffee.
Author
- J. David Creswell
Professor of Psychology and Neuroscience, Carnegie Mellon University
I direct the Center for Modern Digital Health at Carnegie Mellon University in Pittsburgh, where my team and I study how to use wearables to improve health and performance. My lab’s research has relied on consumer devices for years. I wear them for both science and pleasure, and I also wonder how much of what they tell me is true.
Apple’s newest watch , announced Sept. 9, 2026, greets its wearers each morning with a single “readiness” number, built partly from how it thinks they slept. Apple says the redesigned sensors deliver the most accurate heart-rate tracking of any wearable, but independent labs have not yet confirmed that.
Given the growing popularity of sleep trackers, I think it’s worth asking whether a sensor on your wrist or finger can really tell you how your brain was sleeping last night.
Here’s what recent studies show when consumer wearables are tested against the gold standard, polysomnography .
Does it know when I was asleep?
Polysomnography, which combines EEG, or electroencephalogram , with other sensors, records brain waves, eye movements and muscle tone while a person sleeps.
On the simplest question – asleep or awake – today’s devices come close. A 2025 meta-analysis of 24 studies and nearly 800 participants found that wearables typically overestimate total sleep time by about 17 minutes compared with polysomnography. In an eight-hour night, that’s an error of under 4%, roughly two taps of the snooze button.
That gap comes from a blind spot every wearable shares: quiet wakefulness. When you lie awake motionless in the dark, your tracker often mistakes it for sleep. If you sleep well, that rarely matters, because you don’t spend much time lying awake. But if you have insomnia, those long, still stretches add up, and your wearable is probably overestimating how much sleep you’re getting.
Does it know what my brain was doing?
I will start with the obvious: That wearable on your arm or finger does not measure your brain. It measures your heart rate, the beat-to-beat variability of that heart rate, your movement and your skin temperature. The bet the wearable companies are making is that the different sleep stages we go through at night leave trace fingerprints on all of these – REM sleep , for instance, has a distinctive cardiovascular profile – and the wearable’s algorithm reads those sleep fingerprints and makes an educated guess about which stage you are in.
How good is that guess? A 2025 study of six wrist devices worn by 62 adults found only modest agreement with polysomnography on sleep stages, along with significant gaps on time awake after falling asleep and on sleep efficiency. The authors concluded the better-performing watches could track “prolonged and significant changes in sleep architecture.”
Translation: Your deep-sleep number is a noisy nightly estimate, best used to identify significant changes in your sleep over time.
What trackers get right
One of my lab’s findings comes from using these devices for what they’re good at. In a 2023 study in Proceedings of the National Academy of Sciences, we used Fitbits to track the sleep of more than 600 first-year students at three universities, including Carnegie Mellon University. We found that a student sleeping six or fewer hours a night early in the semester had a lower end of term grade-point average relative to their previous semester – an effect that held even after we accounted for academic load and daytime naps.
Nightly sleep duration is the number these devices track well, and it’s the one my students can actually do something about, mostly by getting to bed earlier.
In a 2023 randomized trial , 113 adults with insomnia symptoms spent five weeks in one of two groups. One group got feedback on their sleep as measured by a wearable sensor, along with guidance on how to interpret it. The other got standard sleep-hygiene education: general advice on habits such as keeping a regular bedtime, cutting back on caffeine and screens, and keeping the bedroom dark and quiet.
The feedback group ended up reporting milder insomnia symptoms than the sleep-hygiene group, but the measured amount of sleep they were getting was unchanged. They weren’t sleeping longer, falling asleep faster or waking less. The researchers expected the data to correct a common pattern in insomnia: People often report sleeping less than the sensors record. The feedback didn’t shrink that gap any more than the sleep-hygiene sessions did. What changed was how troubled people felt about their sleep: The feedback group ended up less troubled by it, even though the sleep itself stayed the same.
That distinction between measured sleep and felt sleep is one that sleep scientists here in Pittsburgh have shaped for decades. Researchers at the University of Pittsburgh developed the Pittsburgh Sleep Quality Index in 1989, a questionnaire that asks people to rate how they slept rather than measuring it with wearables. It remains one of the most widely used sleep measures in the world.
The trial’s results suggest that for many people, a wearable’s biggest effect is on the score that the Pittsburgh Sleep Quality Index measures: subjective feelings of one’s sleep quality.
A 2026 trial of 229 older adults in Japan points the same way . Everyone got regular reports from a sleep monitor placed under their bedding. Participants who got the reports alone saw no improvement in their insomnia scores, but those who also got monthly phone guidance on sleep habits and daytime activity did, at least over the first three months. As in the first study, what improved was how people rated their sleep.
A wearable can tell you how much you slept; it can’t make you sleep more. That part is up to you, or to someone who helps you.
And for some people, the tracking itself becomes the problem. Sleep clinicians have a name for patients who lie awake worrying about their sleep score: orthosomnia , a term coined in 2017 and still very much in use . If checking your score reliably wrecks your morning, the device isn’t working for you.
How I read my own numbers
I read the sleep stages and readiness scores the way I read a weather forecast: a useful trend but a poor guide to any single night. And I keep in mind that by the time a study comes out, the device it tested has often been replaced .
Read more of our stories about Pittsburgh and Pennsylvania .
