Sleep Measurement Using Wrist-Worn Accelerometer Data Compared with Polysomnography
John D Chase1, Michael A Busa2, John W Staudenmayer3
1Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA 01003, USA.
Sensors (Basel, Switzerland)
|July 9, 2022
Summary
Different sleep onset definitions did not affect accelerometer sleep estimates. Wrist-worn accelerometers (AG) showed high agreement but low specificity for sleep-wake detection compared to polysomnography (PSG).
Area of Science:
- Sleep Science
- Biomedical Engineering
- Wearable Technology
Background:
- Polysomnography (PSG) is the gold standard for sleep assessment.
- Accelerometer-based devices offer a more accessible method for sleep monitoring.
- Standardizing sleep onset (SO) definitions is crucial for accurate accelerometer data interpretation.
Purpose of the Study:
- To evaluate the impact of varying sleep onset (SO) definitions on accelerometer-derived sleep estimates.
- To compare accelerometer (AG) sleep metrics against polysomnography (PSG) using different SO rules.
- To assess the agreement and accuracy of AG sleep-wake detection versus PSG.
Main Methods:
- Nineteen participants underwent 48-hour monitoring in a home simulation lab.
- Sleep characteristics were measured using PSG and a wrist-worn ActiGraph GT3X+ (AG).
- AG sleep onset was defined using 1-, 5-, and 10-minute consecutive sleep epochs; PSG used the first 'sleep' score.
Main Results:
- Accelerometer-based sleep-wake detection showed high sensitivity (97.2%) and agreement (89.0-89.5%), but low specificity (23.6-25.1%).
- No significant effect of different sleep onset rules on AG sleep estimates was observed.
- Accelerometers underestimated sleep onset latency (SOL) and wake after sleep onset (WASO), while overestimating total sleep time (TST) and sleep efficiency (SE).
Conclusions:
- Alternative sleep onset definitions do not significantly alter accelerometer-derived sleep metrics compared to PSG.
- Current accelerometer algorithms require refinement, potentially by integrating biometric signals like heart rate, to improve accuracy.
- Further research is needed to enhance sleep-wake detection algorithms for wearable devices.
