1119 Validation of a Portable Sleep Electroencephalography Device in Good Sleepers and People with Sleep Apnea

Introduction

Sleep wearable restricted to accelerometry or heart rate monitoring have proved to be helpful to delineate sleep-wake profiles outside of the laboratory environment, but have limited accuracy and only provide indirect estimations of sleep and wake states. This study aimed to assess the validity of a novel electroencephalography headband for ambulatory sleep monitoring as compared to standard polysomnography in good sleepers and individuals with sleep apnea.

Methods

Forty-seven adult males and females from the community took part to this study. This includes a preliminary sample of eight individuals with sleep apnea detected through level 1 polysomnography. All participants underwent one night of in-laboratory sleep recording with the portable EEG headband (MUSE-S, Interaxon) and simultaneous standard polysomnography (Embla N7000/RemLogic, Natus). The Muse-S headband is a commercially available consumer headband with 7 EEG sensors: 2 on the forehead, 2 behind the ears, and 3 reference sensors. MUSE-S data was scored using an automated sleep staging algorithm. Polysomnography data was scored by an independent registered technologist who was blinded to the MUSE-S algorithm-based scoring.

Results

In the overall sample, the accuracy of the Muse-S relative to standard polysomnography ranged between 88% and 96% across all sleep stages, with a sensitivity of 79% to 92%, and a specificity of 90% to 99%. Cohen’s Kappa for all sleep stages combined was 0.76 (CI:0.75-0.76). Analyses per sleep stages showed that Cohen’s Kappa scores were in the fair agreement range for NREM 1 sleep (K=0.41, CI:0.39-0.43), increased to the substantial agreement range for both NREM 2 (K=0.75, CI:0.74-0.75) and NREM3 sleep (K=0.77, CI:0.76-0.77), and further increased to the near perfect agreement range for REM sleep (K=0.85, CI:0.85-0.86) and wake (K=0.84, CI: 0.83-0.84). Similar results were obtained in the subgroup with sleep apnea (overall K= 0.87, CI:0.85-0.88; NREM 1 K=0.33, CI:0.28-0.38; NREM 2 K=0.72, CI:0.70-0.73; NREM3 K=0.81, CI:0.79-0.79; REM K=0.80, CI:0.78-0.82) and wake (K=0.86, CI:0.85-0.88).

Conclusion

Portable EEG-based sleep monitoring with the MUSE-S shows good validity for sleep macroarchitecture variables relative to standard polysomnography. Fair to near perfect concordance was observed across sleep stages in a diverse sample of good sleepers and people with sleep disorders.

Support (if any)