Quantifying individualized temporal structure of neurophysiological and respiratory sleep events: a point process approach
OA Version
Citation
Abstract
Sleep is a multifaceted, dynamic process essential for brain health and cognition, supporting functions such as memory consolidation, synaptic plasticity, neuronal homeostasis, and circadian rhythm regulation. During sleep, distinct waveforms emerge in electrophysiological and respiratory signals, reflecting critical events that capture key aspects of sleep neurophysiology. Changes in sleep events have been closely linked to the progression of Alzheimer’s disease, aging, and other neurological and systemic disorders. Thus, by understanding how sleep events evolve in relation to neurophysiological processes and conditions, we can inform early detection, diagnosis, and future interventions for pathological aging and disease. Despite the dynamic nature of sleep events, most studies rely on static descriptive statistics, ignoring event temporal patterns. The timing and patterns of events are important for shaping neural plasticity during sleep and maintaining sleep stability by modulating the brain's responsiveness to external and internal stimuli. To address this gap, we apply and adapt an integrated approach based on point processes to quantify the moment-to-moment regulation of sleep events, as well as multiple influencing factors and their interactions. Remarkably, we revealed that respiratory events (apneas) and neurophysiological events (spindles) share uniquely individualized timing patterns despite being fundamentally different phenomena. In Chapter 2, we examined obstructive sleep apnea (OSA), a condition of stopped or shallow breathing during sleep that affects ~ 30 million Americans. OSA is clinically measured by the apnea-hypopnea index (AHI), which collapses apnea dynamics into an average event rate. To recover the lost dynamics, we estimated an “instantaneous AHI” as a function of body position, sleep stage, and past events. We revealed individualized respiratory event patterns, with a refractory period followed by an increased propensity period. In Chapter 3, we studied another type of sleep event—sleep spindles, which are ~10-16 Hz brief bursts considered critical for sleep stability, memory, and aging. By modeling multiple factors on moment-by-moment spindle production, we showed fingerprint-like spindle timing patterns, with a refractory period followed by an excitatory period, stable across nights yet heterogeneous across individuals. Crucially, we discovered that temporal patterns are the dominant drivers of event generation. In OSA, incorporating event history dramatically improved model predictability by orders of magnitude. For spindles, short-term (<15 s) timing patterns explain over 70% of the statistical deviance, surpassing other influencing factors like sleep depth and cortical up/down states. Moreover, we observed significant timing pattern differences across age, gender, and racial/ethnic groups. Together, these results set the foundation for a rigorous framework for studying sleep events and their influencing processes across individuals and populations, paving the way for advancements in sleep biomarker development for health and disease.
Description
2025
License
Attribution-NonCommercial-NoDerivatives 4.0 International