Assessment of a wireless headband for automatic sleep scoring

Abstract

Purpose: Classically, professional assessment of sleep is done in the sleep laboratory using whole-night polysomnography (PSG). However, given a misbalance between accredited sleep laboratories and the large amount of patients suffering from sleep disorders, only few receive appropriate diagnostic assessment. Recently, some low-cost home sleep scoring systems have been proposed, yet such systems are rarely tested scientifically. The aim of the present study was to evaluate the staging accuracy of the home sleep scoring system Zeo (Newton, MA, USA). Methods: A final sample of 21 nights from ten subjects (aged 23-45) was digitally recorded with PSG as well as with the Zeo system. We compared scorings of Zeo (on an epoch-be-epoch basis) with the Somnolyzer 24 × 7 (an automatic staging algorithm), expert scorers as well as the freeware SleepExplorer. Results: It was revealed that Zeo shows moderate overall agreement as compared to our study standard Somnolyzer 24 × 7 (κ = 0.56). The most obvious performance difference between Zeo and both other scoring approaches was stage wake (sleep onset latency + wake after sleep onset). While Zeo detected only 40.8 % of the study standard wake epochs, 70.1 % were detected by the expert scorers and 83.4 % by the SleepExplorer, respectively. Conclusions: Data suggest that the Zeo system produces acceptable sleep scoring for stage REM, light and deep sleep, with a specific weakness in correctly detecting waking periods. © 2012 The Author(s).