Multinomial Markov-chain model of sleep architecture in Phase Advanced Subjects

The phase advanced sleep model is used to induce transient insomnia, where subjects go to sleep several hours before their usual bedtime disrupting their normal sleep architecture. The ability of a drug to allow a subject to sleep during this otherwise normal wake time may predict efficacy in insomnia patients. The aims of this work were to: (1) model sleep stage transition probabilities from polysomnography data (PSG) in phase advanced subjects (PAS) over 13 hours after placebo administration, and (2) compare the transition probabilities in PAS to insomniac patients to identify differences in sleep architecture between these two populations for the first 8-hours. Transition probabilities for PAS from two placebo-controlled, parallel studies at two different sites were modeled using a recently reported mixed-effect Markov-chain model based on transition probabilities as multinomial logistic functions in insomniac patients examined after placebo dosing. The multinomial Markov-chain model robustly described phase advanced sleep over the 13-hour observation period after placebo dosing. Compared to insomniac patients, PAS generally displayed lower transition frequencies, fell asleep quicker, spent less total time in REM and ST2, displayed a higher tendency to awaken during early portions of the night and generally, displayed different sleep architectures. The multinomial mixed-effect Markov-chain model provides a useful tool for analyzing sleep data in PAS and may therefore prove useful in the analysis of PSG data from clinical sleep studies investigating sleep promoting drugs using the phase advanced model as a surrogate for efficacy in insomniac patients.