Selecting a sleep tracker from EEG-based, iteratively improved, low-cost multisensor, and actigraphy-only devices.

Area of Science:

• Sleep Science
• Wearable Technology
• Biomedical Engineering

Background:

• Wearable sleep trackers are increasingly used for sleep monitoring.
• Evaluating their accuracy against polysomnography (PSG) is crucial for research applications.
• Different tracker classes, from EEG-based to low-cost consumer devices, vary in their technological sophistication.

Purpose of the Study:

• To assess the performance of six distinct wearable sleep trackers.
• To compare tracker accuracy across different sleep classification stages (2-stage and 4-stage).
• To provide recommendations for selecting appropriate trackers based on study needs.

Main Methods:

• Six wearable devices were evaluated: Dreem 3 (EEG), Actigraph GT9X (actigraphy), Oura Ring Gen3, Fitbit Sense, Xiaomi Mi Band 7, and Axtro Fit3 (consumer trackers).
• Sixty participants across three age groups slept overnight in a laboratory setting.
• Epoch-by-epoch and discrepancy analyses were conducted comparing device data against in-lab polysomnography (PSG) with consensus scoring.

Main Results:

• The EEG-based Dreem headband demonstrated the highest accuracy (2-stage kappa=0.76, 4-stage kappa=0.76-0.86).
• Iteratively improved trackers (Oura, Fitbit) showed moderate accuracy (2-stage kappa 0.47-0.64), outperforming basic actigraphy.
• Low-cost trackers exhibited the poorest performance (2-stage kappa <0.31).
• Discrepancies were larger during nights with poorer sleep quality.

Conclusions:

• The EEG-based Dreem headband is recommended for studies requiring high sleep-staging accuracy or evaluating sleep disturbances.
• Moderately accurate consumer trackers (Oura, Fitbit) offer a balance of performance and usability for general research on healthy sleepers.
• Low-cost trackers are suitable only for basic time-in-bed logging and not for research-grade sleep analysis.