ABSTRACT
Arousals during sleep are transient accelerations of the EEG signal, considered to reflect sleep perturbations associated with poorer sleep quality. They are typically detected by visual inspection, which is time consuming, subjective, and prevents good comparability across scorers, studies and research centres. We developed a fully automatic algorithm which aims at detecting artefact and arousal events in whole-night EEG recordings, based on time-frequency analysis with adapted thresholds derived from individual data. We ran an automated detection of arousals over 35 sleep EEG recordings in healthy young and older individuals and compared it against human visual detection from two research centres with the aim to evaluate the algorithm performance. Comparison across human scorers revealed a high variability in the number of detected arousals, which was always lower than the number detected automatically. Despite indexing more events, automatic detection showed high agreement with human detection as reflected by its correlation with human raters and very good Cohen's kappa values. Finally, the sex of participants and sleep stage did not influence performance, while age may impact automatic detection, depending on the human rater considered as gold standard. We propose our freely available algorithm as a reliable and time-sparing alternative to visual detection of arousals.
ABSTRACT
BACKGROUND: Obstructive sleep apnea syndrome (OSAS) is common in adults. People with OSAS have a higher risk of experiencing traffic accidents and occupational injuries (OIs). We aimed to clarify the diagnostic performance of a three-channel screening device (ApneaLinkTM) compared with the gold standard of full-night attended polysomnography (PSG) among hospital outpatients not referred for sleep-related symptoms. Furthermore, we aimed to determine whether manual revision of the ApneaLinkTM autoscore enhanced diagnostic performance. METHODS: We investigated 68 patients with OI and 44 without OI recruited from the University Hospital Basel emergency room, using a cross-sectional study design. Participating patients spent one night at home with ApneaLinkTM and within 2 weeks slept for one night at the sleep laboratory. We reanalyzed all ApneaLinkTM data after manual revision. RESULTS: We identified significant correlations between the ApneaLinkTM apnea-hypopnea index (AHI) autoscore and the AHI derived by PSG (r = 0.525; p <0.001) and between the ApneaLinkTM oxygen desaturation index (ODI) autoscore and that derived by PSG (r = 0.722; p <0.001). The ApneaLinkTM autoscore showed a sensitivity and specificity of 82% when comparing AHI ≥5 with the cutoff for AHI and/or ODI ≥15 from PSG. In Bland Altman plots the mean difference between ApneaLinkTM AHI autoscore and PSG was 2.75 with SD ± 8.80 (ß = 0.034), and between ApneaLinkTM AHI revised score and PSG -1.50 with SD ± 9.28 (ß = 0.060). CONCLUSIONS: The ApneaLinkTM autoscore demonstrated good sensitivity and specificity compared with the gold standard (full-night attended PSG). However, Bland Altman plots revealed substantial fluctuations between PSG and ApneaLinkTM AHI autoscore respectively manually revised score. This spread for the AHI from a clinical perspective is large, and therefore the results have to be interpreted with caution. Furthermore, our findings suggest that there is no clinical benefit in manually revising the ApneaLinkTM autoscore.
