(Reuters Health) – Sleep-wake patterns tracked by two popular consumer wearables – the Apple Watch and the Oura Ring – gave results comparable to those found with actigraphy and polysomnography in a small experiment.
Multi-sensor wearable consumer devices allowing collection of multiple data sources, such as heart rate and motion, for the evaluation of sleep in the home environment, are increasingly ubiquitous, researchers note in the journal Sleep. However, the validity of such devices for sleep assessment has not been directly compared to alternatives such as wrist actigraphy or polysomnography (PSG).
“We used data from the wearables along with a machine learning model to develop a sleep-wake classifier using the sensors of relatively cheap consumer wearables like the Apple Watch,” said lead author Daniel Roberts of Sonic Sleep in New York City.
“Our findings add to the understanding of how to leverage these new ‘multisensor’ wearable devices to accurately measure sleep, which could be useful in the future to detect disordered sleep outside of the lab,” Roberts said by email.
For the current analysis, each of eight participants completed four nights in a sleep laboratory, equipped with (PSG) and several wearable devices. RPSGT-scored PSG served as ground truth for sleep-wake state. Wearable devices providing sleep-wake classification data were compared to PSG at an epoch-by-epoch level, typically 30-second epochs, and at a full-night level.
Data from multi-sensor wearables (Apple Watch and Oura Ring) were compared to data available from ECG and a tri-axial wrist actigraph to evaluate quality and utility of heart rate and motion data.
Then, machine learning methods were used to train and test sleep-wake classifiers, using data from consumer wearables, and researchers found that data from multi-sensor wearables were strongly correlated at an epoch-by-epoch level with the reference sources.
The study included a relatively small sample of eight healthy participants, all in mid-life, sleeping in a highly controlled laboratory environment. More research is needed to determine how well sleep-wake models trained on healthy participants work for those with sleep disorders and in outpatient settings, the authors note.
Another limitation is that the models were developed in the study over the course of a night, as opposed to over a 24-hour period. A model developed on 24-hour data which contains a greater number of periods of wakefulness may be useful to decrease the bias towards classification of sleep that is common in sleep-wake classification, the researchers point out.
Outside of a laboratory, user error is also possible and could make consumer wearables less accurate for tracking sleep, said Kelly Evenson, a professor at the Gillings School of Global Public Health at the University of North Carolina Chapel Hill.
“It is very important to wear activity trackers as designed by the company,” Evenson, who wasn’t involved in the study, said by email.
“Many people ignore simple set-up functions that would improve the accuracy for their own monitoring,” Evenson said. “For example, does the activity tracker offer a sleep setting (or sleep mode setting) that should be turned on when they are going to bed?”
SOURCE: bit.ly/3b8Xu3R Sleep, online March 26, 2020.