Wearables vs. Polysomnography: The Standard
When it comes to precisely understanding our sleep, Polysomnography (PSG) stands as the undisputed gold standard. This comprehensive clinical assessment involves an overnight stay in a specialized sleep laboratory, where a range of physiological signals are recorded. This includes brain activity (via electroencephalography), eye movements (electrooculography), and muscle tone (electromyography), alongside audio and video recordings. These detailed measurements facilitate accurate classification of sleep into distinct stages: wakefulness, light sleep (N1 and N2), deep sleep (N3), and Rapid Eye Movement (REM) sleep, typically in 30-second intervals known as epochs. This level of detail is crucial for diagnosing various sleep disorders and understanding the nuances of an individual's sleep architecture.
However, despite its unparalleled accuracy, PSG has practical limitations. Its high cost, labor-intensive nature, and the requirement for specialized equipment and medical expertise make it impractical for routine, long-term monitoring in a home environment. An overnight stay in an unfamiliar clinic can also disrupt natural sleep patterns, leading to what's known as the "first-night effect."
This is where consumer wearable sleep-tracking devices have emerged as an appealing alternative, offering users insights into sleep patterns outside the clinical setting, with detailed data in their associated apps. Wearables typically rely on a combination of accelerometers and photoplethysmography (PPG) sensors to gather information. Accelerometers detect body movements throughout the night, helping to differentiate between restful wakefulness and various sleep stages based on movement variations. PPG sensors, on the other hand, use light to measure changes in blood volume under the skin, providing data on heart rate, heart rate variability, and blood flow—physiological markers that vary across sleep stages. By combining these data points, wearables attempt to infer sleep stages, offering a more comprehensive picture than movement alone. However, it's important to note that PPG readings can be influenced by factors like motion artifacts, skin pigmentation, and ambient light.
Given the fundamental differences in how PSG and wearables collect and interpret sleep data, one may ask: how accurate are these consumer devices compared to the gold standard? This was precisely the aim of a recent laboratory-based study by Schyvens et al. [1], which sought to provide an updated evaluation of six popular wrist-worn wearable sleep trackers against PSG, comparing their performance in assessing various sleep parameters.
What did Schyvens' team find?
This comprehensive study aimed to evaluate the performance of six popular consumer wrist-worn wearable devices—the Fitbit Charge 5, Fitbit Sense, Withings Scanwatch, Garmin Vivosmart 4, Whoop 4.0, and Apple Watch Series 8—against the gold standard, PSG, for detecting various sleep parameters. Sixty-two adults (52 males, 10 females, with a mean age of 46 years) participated, including both healthy individuals and those with suspected sleep apnea. Each participant spent a single night in a sleep laboratory, simultaneously undergoing PSG while wearing two to four of the test devices.
The researchers conducted an epoch-by-epoch analysis, comparing 30-second intervals of sleep data from the wearables to those derived from PSG. This allowed for a detailed assessment of two-state categorization (sleep vs. wake) and four-state categorization (wake, light sleep, deep sleep, and rapid eye movement, or REM sleep).
Here's what the study uncovered about their performance:
- Overall sleep detection: All devices demonstrated high sensitivity, correctly identifying over 90% of sleep epochs. However, their specificity for detecting wake periods was considerably lower, ranging from 29.39% to 52.15%, meaning they frequently misclassified wakefulness as sleep.
- Sleep time & efficiency: Most wearables tended to overestimate the total amount of time spent asleep by 6.31 to 39.87 minutes, with the Fitbit Charge 5 being an exception, slightly underestimating by 5.74 minutes. Similarly, all devices significantly overestimated the percentage of time in bed spent asleep by 2.20% to 10.19%. While some overestimations might be clinically acceptable for general wellness, larger discrepancies could be misleading.
- Sleep stages: While devices generally measured deep sleep and REM sleep more accurately than wake and light sleep, significant under- or overestimations were still observed. A common error across all wearables was the misclassification of wake, deep sleep, and REM sleep epochs as light sleep. The Apple Watch Series 8 showed the highest accuracy for identifying wake (52.15%) and REM sleep (68.57%), while the Whoop 4.0 performed best for deep sleep (69.63%).
- Agreement with PSG: The research team used a scoring system from 0 to 1 to quantify how closely the wearables match the PSG measurements. A score of κ = 1 means perfect agreement, and 0 means no better than random guessing. Trained PSG technicians typically score κ = 0.75 amongst themselves. The wearable scores ranged from 0.21 to 0.53, indicating fair to moderate agreement with PSG. The Apple Watch Series 8 (κ = 0.53), Fitbit Sense (κ = 0.42), and Fitbit Charge 5 (κ = 0.41) showed the highest agreement.
- Wrist Position: The study also found that the performance of the wearables was largely consistent regardless of which wrist or position they were worn on, with only minor, non-substantial variations.
In essence, while the study revealed varying degrees of accuracy across devices for different sleep parameters, it highlighted that some, like the Fitbit Sense, Fitbit Charge 5, and Apple Watch Series 8, showed more promising results for overall sleep duration and efficiency, despite challenges in precise sleep stage differentiation and wake detection.
What This Means for You
The bottom line is straightforward: your sleep tracker is a useful wellness tool, not a medical device. Think of it like a bathroom scale—helpful for spotting trends over time, but not a substitute for a doctor's assessment. These devices do a reasonable job tracking your overall sleep duration across nights, which makes them genuinely useful for noticing patterns. Are you consistently short on sleep? Does your rest improve when you cut evening caffeine or stick to a regular bedtime? Wearables can help answer these questions and motivate better sleep habits through daily feedback. This study found the Apple Watch Series 8 performed best, followed by the Fitbit Sense and Fitbit Charge 5. The Whoop 4.0, Withings Scanwatch, and Garmin Vivosmart 4 lagged behind.
Fitness trackers do struggle in important ways. Wearables frequently miss brief nighttime awakenings and lack precision when categorizing sleep stages—that "90 minutes of deep sleep" your app reports may be off by 30 minutes or more. They also cannot reliably identify sleep disorders, like insomnia or sleep apnea—only a clinical sleep study can do that.
Instead, use your data wisely by focusing on week-to-week trends rather than obsessing over nightly numbers. A steady decline in sleep duration over a month is meaningful and worth addressing. A single night's disappointing "deep sleep" reading is not—the margin of error is simply too high for that level of precision.References
- Schyvens AM et al. "A performance validation of six commercial wrist-worn wearable sleep-tracking devices for sleep stage scoring compared to polysomnography", Sleep Advances 6 (2025) zpaf021. https://doi.org/10.1093/sleepadvances/zpaf021