Abstract Rationale During sleep, occurrence of obstructive events varies across sleep stages but their precise timing within a given sleep stage is hard to predict. Yet, sleep stages are not homogenous and within-stage modulations of cortical activation can be tracked with the analysis of EEG dynamics. We hypothesized that obstructive event occurrence depends on the level of cortical activation assessed by surface EEG. Methods We conducted a monocentric retrospective study including patients who underwent overnight video-polysomnography in our sleep department for the diagnosis of obstructive sleep apnea between July 2022 and February 2023. EEG signals (F3, C3, O1 referenced to A2; 256 Hz) were filtered (0.5-45 Hz, 50/100 Hz notch). For each 30-second epoch we measured Kolmogorov complexity, a proxy for cortical activation, over 10-second segments. For epochs with an obstructive event (apnea and hypopnea), we analyzed the 10-second segment preceding the event. From epochs with no obstructive event, we randomly sampled one 10-second control segment matching the start-time distribution observed for pre-event segments. Within each stage and for each patient, we split segments at the median Kolmogorov value. Epochs with a Kolmogorov complexity above median were classified as having a high level of cortical activation and others with a low level. Proportion of epochs with or with no obstructive events were compared using a chi-square test. Results We included 50 patients (50% women, mean age: 56.38 ± 15.21, mean apnea-hypopnea index of 29.9 events per hour 19.02 - 45.33, sleep duration of 428.75 minutes 353.12 - 449.00). We included 28,679 epochs in our analysis, 8,015 (27.9%) of them contained obstructive events. Prevalence of obstructive events was higher in epochs with a high level of cortical activation as compared to those with a low level: 30.2% vs 25.7% (p 0.001). This difference was mainly driven by N2 and N3 differences (figure 1). The association between complexity and presence of obstructive event did not differ by sex, age (≤57.5 vs 57.5 y), or Epworth (≤9 vs 9) in any stage (all interaction p ≥ 0.37). Conclusions Higher EEG-derived cortical activation is associated to higher obstructive event occurrence, implying that for constant stage and body position, events track brain state within NREM sleep predominantly. This abstract is funded by: None
Raoul-Duval et al. (Fri,) studied this question.