PSG Validation of minute-to-minute scoring for sleep and wake periods in a consumer wearable device.
Level 3 - non-randomized controlled study
Diagnostic validation study comparing consumer and research-grade actigraphy against reference-standard polysomnography.
PubMed 32941498 · doi:10.1371/journal.pone.0238464
What was done
Sleep clinic patients (n = 41) wore a consumer-grade wearable (Huami Arc) on their non-dominant wrist during overnight polysomnography (PSG). The sample was split into a training group (n = 20) to calibrate threshold settings and a testing group (n = 21) that also simultaneously wore a research-grade actigraph (Philips Actiwatch Spectrum). Epochs of 60 seconds were scored for sleep and wake on both devices using the Cole-Kripke algorithm and compared against PSG.
What was found
In the training group, the optimal threshold for the Arc device was 10 units. In the testing group, the Arc had an accuracy of 90.3% ± 4.3%, sleep sensitivity of 95.5% ± 3.5%, and sleep specificity of 55.6% ± 22.7%. The research-grade Actiwatch had an accuracy of 88.7% ± 4.5%, sleep sensitivity of 92.6% ± 5.2%, and sleep specificity of 60.5% ± 20.2%.
Why it matters
A calibrated consumer-grade wearable can perform comparably to expensive research-grade actigraphs for epoch-by-epoch sleep estimation.
Limits
The sample was small (41 total, 21 in testing) and limited to sleep clinic patients evaluated over a single night. Both devices showed poor specificity for sleep (55.6% and 60.5%), reflecting actigraphy's standard limitation in distinguishing motionless wakefulness from sleep. Additional data sources on the device like photoplethysmography were not tested.
Cited by
- context Wrist-based movement detection can determine whether someone is awake or asleep minute-to-minute with over 90% accuracy relative to EEG brain wave activity.