THE ODDS RATIO PRODUCT AS A MARKER OF SLEEP HOMEOSTASIS

Abstract Introduction Classic EEG markers of sleep homeostasis (process S) usually study the level and amount of slow wave activity (SWA). In recent years, the odds ratio product (ORP) has been described as a novel quantitative measure of sleep depth (and conversely, arousability), which is derived from a four-digit permutation of the power in four respective frequency bands. Therefore, we sought to determine whether the ORP can also be used as a novel quantitative marker of process S. Methods Utilizing C3/C4 electrodes with TP9/TP10 reference, ORP was calculated from 25 nights in 15 patients. We used ORP from REM and NREM1-3 segments/bouts in each person’s night(s) with varying number of data points per REM segment(s) in each night(s). Artifacts and wake segments were discarded from analysis. Average ORP is calculated for an individual bout per night per patient. Generalized Estimating Equation (GEE) modelling used GEEQBOX MATLAB toolbox to identify change in sleep depth over bouts. Resultant beta (β) values represented change in nightly sleep depth over sequential NREM/REM segments per patient. Significance was p≤0.000833. Results From 15 people in the epilepsy monitoring unit, 3 did not have epilepsy. In NREM, the average ORP for C3/C4 is 1.26/1.19 (n=146307/125709). Average number of nightly NREM segments for C3/C4 are 4.64/4.9 (n=1261.27/1282.74 average ORP values per nightly segment). In REM, the average ORP for C3/C4 is 1.44/1.44 (n=47629/43528). Average number of nightly REM segments for C3/C4 are 4.36/4.38 (n=418.68/453.33 average ORP values per nightly segment). Signifying the change in sleep depth the beta values from C3/C4 were consistently positive: REM (epilepsy 0.033/0.018, non-epilepsy 0.096/0.053), NREM (epilepsy 0.047/0.053, non-epilepsy 0.026/0.064). Conclusion As a marker of sleep homeostasis, SWA decreases across a period of sleep. Consistently positive beta values derived from ORP in both REM and NREM are comparable because they demonstrate shallower sleep depth (i.e. greater arousability) across segments over nights as sleep debt is repaid. Also as expected, REM has a higher ORP than NREM. Furthermore, our findings of positive beta values are robust against sleep stage and pathology (i.e. epilepsy). Therefore, these findings may represent a novel quantitative marker of sleep homeostasis that includes REM. Support (If Any)