Sleep Trackers:
How They Work, How Accurate They Are and How to Use Your Sleep Data

Millions of people now track their sleep with smartwatches, rings, fitness bands, under-mattress pads and phone apps. These devices can be useful, but it's important to understand what they can and can't measure. The gold standard for measuring sleep is polysomnography (PSG), an overnight test in a sleep laboratory that records brain waves, eye movements, muscle activity, breathing and oxygen levels. Consumer trackers don't measure brain activity. Instead, they estimate sleep from movement, heart rate, heart rate variability, breathing and skin temperature, using each company's own algorithm. The evidence is consistent: modern trackers are good at telling when you're asleep, with more than 90% of sleep periods detected correctly, and reasonably good at estimating how long you slept. They're much weaker at detecting when you're awake in bed, and at telling which sleep stage you're in. In a 2025 Belgian study of six popular devices against PSG, all detected sleep well but correctly identified only 29-52% of time spent awake, overestimated total sleep by about 6 to 40 minutes, and agreed with PSG on sleep stages only fairly to moderately. In a 2022 Australian study of six wearables, sleep-stage agreement ranged from 50% to 65%. No consumer sleep tracker can diagnose a sleep disorder, although the Apple Watch and Samsung Galaxy Watch now have regulator-authorised features that can flag possible sleep apnoea. Used well, sleep trackers help you see patterns, keep a regular schedule and spot changes over time. Used badly, they can cause anxiety about sleep, a problem doctors call "orthosomnia". This article explains how sleep trackers work, how accurate each type is, and how to use your data sensibly.

Key numbers

How sleep is measured properly

Polysomnography. In a sleep laboratory, polysomnography records:

• Brain waves (EEG): the only reliable way to identify sleep stages.
• Eye movements: to identify REM (rapid eye movement) sleep.
• Muscle activity, heart rhythm, breathing, blood oxygen and leg movements.

Trained technicians score every 30 seconds of the night into one of five states (see Sleep Architecture):

• Wake.
• N1 and N2 (light sleep): about half or more of the night.
• N3 (deep or slow-wave sleep): important for physical recovery and memory, and more common early in the night.
• REM sleep: when most dreaming occurs, important for emotional processing and memory, and more common later in the night (see Sleep and Memory Consolidation).

Other clinical tools.

• Home sleep apnoea tests: record breathing, oxygen and heart rate to diagnose obstructive sleep apnoea at home.
• Actigraphy: a research-grade wrist device that estimates sleep from movement over days or weeks.
• Sleep diaries: simple daily records of bedtime, wake time and how you slept, still widely used in insomnia treatment.

How consumer sleep trackers work

Consumer trackers estimate sleep from signals they can measure, then use machine-learning algorithms trained on PSG data:

• Movement (accelerometer): you move less when asleep. Early trackers used only movement, so they often mistook lying still and awake for sleep.
• Heart rate: falls during sleep, especially deep sleep, and becomes more variable during REM.
• Heart rate variability (HRV): changes with sleep stages and the balance of the nervous system (see Heart Rate Variability).
• Breathing rate: estimated from heart rate patterns or movement.
• Skin temperature: used by rings and some watches to detect changes over nights, such as during illness or the menstrual cycle.
• Blood oxygen (SpO2): some devices estimate it overnight, and changes may suggest breathing problems.
• Sound: some phone apps and smart speakers listen for snoring and breathing.
• Radar or pressure sensors: "nearables" under the mattress or on the bedside table detect movement and breathing without being worn.

Because no consumer device measures brain waves (except some specialised EEG headbands), sleep stages are estimated, not measured.

Types of sleep trackers

What the evidence shows

Good at detecting sleep, poor at detecting wake. In a 2025 study in Sleep Advances, An-Marie Schyvens and colleagues at Antwerp University Hospital compared six devices with PSG in 62 adults, including people with suspected sleep apnoea: the Apple Watch Series 8, Fitbit Sense, Fitbit Charge 5, WHOOP 4.0, Withings ScanWatch and Garmin Vivosmart 4.

• Sleep detection: all devices correctly identified 92-96% of sleep periods.
• Wake detection: they correctly identified only 29-52% of awake periods, so time spent lying awake was often counted as sleep.
• Total sleep time: most devices overestimated it by about 6 to 40 minutes; the two Fitbits' differences weren't statistically significant.
• Time awake during the night: underestimated by about 12 to 48 minutes.
• Sleep stages: agreement with PSG was fair to moderate (kappa 0.21-0.53). The Apple Watch, Fitbit Sense and Fitbit Charge 5 performed best; the Withings ScanWatch and Garmin Vivosmart 4 performed worst.
• Conclusion: the authors judged the Fitbit Sense, Fitbit Charge 5 and Apple Watch Series 8 clinically acceptable for total sleep time and sleep efficiency, but stated that "wearables do not replace PSG for clinical diagnosis."

The study tested one night per person, and most participants were men with lighter skin.

Sleep stages are the weakest point. In a 2022 study in Sensors, Dean Miller and colleagues at Central Queensland University compared six devices with PSG in 53 healthy young adults: WHOOP 3.0, Oura Ring Gen 2, Apple Watch Series 6, Garmin Forerunner 245, Polar Vantage V and Somfit (a forehead device). For sleep versus wake, all devices agreed with PSG 86-89% of the time. For sleep stages, agreement was lower: 65% for Somfit, 61% for Oura, 60% for WHOOP, 53% for Apple Watch, 51% for Polar and 50% for Garmin. The authors concluded that all six were valid for measuring sleep timing and duration but needed improvement for classifying sleep stages.

Rings and newer devices. Oura has published results from several validation studies showing around 76-79% agreement with PSG for four sleep stages and about 92-96% for sleep versus wake, but these are company-reported summaries, and independent studies have generally found lower agreement for sleep stages. Newer models of most brands are likely to perform better than older ones, but many haven't yet been independently tested.

The AASM position. The American Academy of Sleep Medicine's 2018 position statement concluded that consumer sleep technologies "cannot be utilized for the diagnosis and/or treatment of sleep disorders at this time", but may help conversations between patients and doctors when used within a proper clinical assessment.

Sleep apnoea features

Obstructive sleep apnoea, in which the airway repeatedly narrows or closes during sleep, is common and often undiagnosed (see Sleep Apnea). It's linked to high blood pressure, heart disease, stroke, type 2 diabetes, accidents and daytime sleepiness. Two smartwatch features have now been authorised by the US FDA to flag possible sleep apnoea:

• Samsung Galaxy Watch received FDA De Novo authorisation in February 2024 for a feature that can flag signs of possible moderate to severe obstructive sleep apnoea after two nights of at least four hours of sleep tracking within 10 days, for adults aged 22 and over without a diagnosis.
• Apple Watch (Series 9 and later, and Ultra 2) received authorisation in 2024 for sleep apnoea notifications based on "breathing disturbances" detected by the accelerometer over 30 days. In Apple's validation study of 1,448 adults compared with home sleep apnoea tests, the feature correctly identified 66% of people with moderate to severe apnoea (89% of severe cases but only 43% of moderate cases) and correctly ruled it out in 98.5% of those without it.

These features are screening tools, not diagnoses. They miss many people with apnoea, especially moderate cases, so no notification doesn't mean you don't have sleep apnoea. If you snore loudly, stop breathing or gasp during sleep, or feel very sleepy during the day, see a doctor whatever your tracker says.

Sleep scores and readiness scores

Most trackers combine sleep duration, timing, stages, heart rate, HRV and other data into a single "sleep score" or "readiness score". Each company calculates these differently, and the formulas aren't public or standardised. A score of 80 on one device doesn't mean the same as 80 on another. Scores can be motivating, but they're best used as a rough guide to your own trends, not as an objective measure of sleep quality.

The risk of orthosomnia

In 2017, sleep researcher Kelly Glazer Baron and colleagues described patients who became so focused on achieving "perfect" tracker data that it worsened their sleep. They called it orthosomnia. Some patients spent longer in bed trying to raise their sleep totals, which can actually worsen insomnia. One trusted her tracker's verdict of poor sleep over a laboratory test showing she'd slept well. As Baron put it, "They are not able to differentiate between light and deep sleep." If checking your sleep data makes you anxious, take a break from it.

Why this matters for longevity

Sleep is one of the foundations of healthy ageing. Short and irregular sleep are linked to higher risk of heart disease, type 2 diabetes, depression, dementia and early death, and adults need at least seven hours a night (see Sleep Duration and Longevity). Sleep trackers can help by showing how much and how regularly you sleep, and by highlighting the effects of late caffeine, alcohol, screens, travel and stress. They work best as tools for building habits and spotting changes, combined with how you actually feel. They can't replace a medical assessment for sleep problems. The same caution applies to other wearable numbers (see Resting Heart Rate and HRV and Fitness Tracker Calories).

When to see a doctor about your sleep

Regardless of what your tracker shows, see a doctor if you have:

• Loud snoring, pauses in breathing or gasping during sleep.
• Daytime sleepiness, such as falling asleep while reading, watching TV or driving.
• Difficulty falling or staying asleep at least three nights a week for three months or more (see Insomnia).
• Waking unrefreshed despite enough time in bed.
• Unusual behaviours during sleep, such as acting out dreams, or an urge to move your legs at night.
• High blood pressure that's hard to control, which can be linked to sleep apnoea.

Recommendations by situation

Practical notes

Wear your device snugly and consistently, charge it during the day, and keep your profile accurate. Look at weekly averages and trends rather than individual nights, and compare the numbers with how rested you actually feel. Use your data to test changes, such as an earlier bedtime, less evening caffeine or alcohol, or a consistent wake time, and see what helps. Don't spend extra time in bed just to improve your score. If you have symptoms of a sleep disorder, see a doctor rather than relying on your tracker. Our Longevity Doctors review sleep alongside heart, metabolic and brain health in every assessment and can help you interpret your sleep data and arrange further testing if needed, starting with the free longevity assessment.