To assess the use of an automated system (Morpheus) in the analysis of human EEG frequency changes in pre and post severe obstructive sleep apnea using a split night protocol. To correlate adaptive segmentation and fuzzy logic frequency changes using Markov models to traditional R&K.
A total of 23 adults were studied with a diagnosis of severe obstructive sleep apnea. Each subject underwent polysomnography (PSG) and received early intervention CPAP therapy during initial study. Only individuals with oxygen saturations greater than 85% and respiratory disturbance index (RDI) less than 10 episodes per hour during the ideal CPAP titration period were included. Fundamental frequency segments were measured using Morpheus algorhythms. Sleep states were analyzed using traditional R&K rules.
Means and standard deviations are reported for 23 adults. RDI pre CPAP was 73.5 (26) with low oxygen saturation of 81%(7). Pre and post CPAP periods for the sleep states that reached statistical significance by automated analysis during study time were: high frequency(HF)=pre 30%(15), post 23%(11); low frequency(LF)= pre 3%(4), post 8%(5); mixed frequency-1(MF1)=pre 15%(8), post 25%(7); delta= pre 2%(4), post 4%(5); beta= pre 5%(5),post 2%(2); under 4 Hz= pre 5.2%(7), post 12.6%(5). Significant R&K sleep state results are as follows as a % of time in bed: sleep efficiency pre= 76%(16), post 84%(9); Stage II pre= 58%(16), post 35%(12); Stages III&IV pre= 4%(8), post 15(9); REM pre=4%(9), post 26(11). Signals analyzed but not found to be statistically significant were: Morpheus mixed frequency 2, theta, alpha, and sigma segments; R&K Stage I.
The automated analysis demonstrates a different methodogy for analyzing a known biological sleep signal. In this study group with severe sleep apnea there was an improvement in sleep measures that reflect EEG synchrony comparable to that as measured by R&K.
A completely automated analysis may expand knowledge of sleep states and processes from a high resolution multidimensional mathematical perspective thus allowing improved understanding of disease state mechanisms, medical risk, treatment outcomes, and quality of life measures.
Richard Bogan, Shareholder SleepMed Inc.