Validity and clinical utility of a wrist-worn device against polysomnography

Sep 30, 2025PloS one

Accuracy and usefulness of a wrist device for sleep measurement compared to standard sleep testing

AI simplified

Abstract

In a study of 98 participants, the smart watch demonstrated a sensitivity of 95.9% for sleep/wake performance compared to .

  • The smart watch achieved an accuracy of 87.3% and a positive predictive value of 72.2% for sleep/wake detection.
  • Total sleep time was overestimated by the smart watch by an average of 28.7 minutes compared to polysomnography.
  • Sleep efficiency was overestimated by 5.94% and sleep onset latency by 8.53 minutes when compared to polysomnography.
  • Wake after sleep onset was underestimated by 37.00 minutes with the smart watch.
  • Acceptable agreement was found in sleep/wake detection and total sleep time, particularly with Kappa and intraclass correlation coefficients greater than 0.4.
  • Agreement was less satisfactory among participants diagnosed with obstructive sleep apnea or insomnia.

AI simplified

Key numbers

95.9%
Sensitivity
Performance of the smart watch in sleep/wake detection
28.7 min
Total Sleep Time Overestimation
Difference in total sleep time compared to
47.9%
Specificity
Performance of the smart watch in sleep/wake detection

Full Text

What this is

  • This study validates the HUAWEI WATCH GT2, a consumer-grade sleep-tracking device, against () in clinical settings.
  • Participants with suspected sleep disorders completed while wearing the smart watch to assess its accuracy in sleep detection.
  • The results show that while the smart watch has high sensitivity, it also tends to overestimate certain sleep parameters.

Essence

  • The HUAWEI WATCH GT2 demonstrated high sensitivity (95.9%) for sleep detection but had lower specificity (47.9%). It overestimated total sleep time by 28.7 minutes compared to .

Key takeaways

  • The smart watch showed a sensitivity of 95.9% and an accuracy of 87.3% for sleep/wake detection, indicating strong performance in identifying sleep states.
  • Total sleep time was overestimated by 28.7 minutes, and sleep efficiency was overestimated by 5.94%, highlighting discrepancies in the smart watch's performance compared to .
  • Agreement was acceptable for sleep/wake detection (Kappa coefficient > 0.4), but performance was less satisfactory in patients with obstructive sleep apnea (OSA) or insomnia.

Caveats

  • The study was limited to a single center and included only 98 participants, which may affect the generalizability of the findings.
  • The smart watch's performance was less reliable in patients with sleep disorders, indicating that further validation is necessary for clinical use.
  • The exact algorithm used in the smart watch is unknown, and updates to the device may require new validations to ensure consistent accuracy.

Definitions

  • Polysomnography (PSG): A comprehensive test that records brain waves, oxygen levels, heart rate, and breathing, as well as eye and leg movements during sleep.

AI simplified

what lands in your inbox each week:

  • 📚7 fresh studies
  • 📝plain-language summaries
  • direct links to original studies
  • 🏅top journal indicators
  • 📅weekly delivery
  • 🧘‍♂️always free