The microstructure of cortical activity during sleep: Development of an objective, automatic procedure for sleep stage classification

睡眠障碍是最常见的症状之一，特别是在工业国家，有明显的上升趋势。除了对受影响者造成的直接后果外，慢性睡眠障碍还会对社会造成代价高昂的损害，例如通过与疲劳有关的错误和工作事故，以及人类在控制和监测活动中的失误。因此，睡眠障碍在公共卫生方面是一个重要且日益严重的医学问题。自20世纪70年代以来，睡眠实验室的多导睡眠图一直是诊断许多睡眠障碍的金标准。在睡眠实验室中记录的生物信号的评估睡眠分析一方面是非常耗时的，另一方面受到高的内部和内部评级的可靠性。因此，本项目的主要目标是建立一个新的，完全客观化的，因此自动化的睡眠阶段分析的手段的全球模式分析的皮质活动分布的基础上64通道EEG数据。这一分析将基于客观降维方法，基于超统计时间序列分析的贝叶斯方法，基于描述复杂概率分布的灵活的概率模型，以及基于证据的贝叶斯模型选择。这样的方法可以从整体上更深入地理解睡眠的神经生理学，并且可以具有临床实用的益处，即耗时的视觉/手动睡眠阶段分析可以在将来被自动化的，客观的方法，可以提供睡眠宏观和微观结构的时间分辨和动态图像。此外，在新方法的帮助下，阻塞性睡眠呼吸暂停患者的局部睡眠现象的调查可能会导致对白天过度嗜睡的病理生理学的新见解。

Sleep disorders are among the most frequently complainedsymptoms, especially in the industrial nations, with a clear upwardtrend. In addition to the direct consequences for those affected,chronic sleep disorders cause costly damage to society, e.g. throughfatigue-related errors and accidents at work, and human failures incontrol and monitoring activities. Sleep disorders are therefore asignificant and growing medical problem in terms of public health.Since the 1970s, polysomnography in sleep laboratories has been thegold standard in the diagnosis of many sleep disorders. Theevaluation of the biosignals recorded in the sleep laboratory for sleepanalysis is on the one hand extremely time-consuming and on theother hand subject to a high inter- and intra-rater reliability. Therefore,the main goal of the current project is to establish a novel, completelyobjectified and therefore automatable sleep stage analysis by meansof a global pattern analysis of the cortical activity distribution based on64-channel EEG data. This analysis will be based on objectivemethods of dimensional reduction, on a novel Bayesian method ofsuper-statistical time series analysis, on a flexible mathematicalmodel to describe complex probability distributions, and on evidencebasedBayesian model selection.Such a method could lead to a deeper understanding of the neurophysiology of sleep as a whole, and could have the clinical-practical benefit that the time-consuming visual/manual sleep stage analyses could be replaced in the future by an automated, objective method that could provide a time-resolved and dynamic image of the sleep macro- and microstructure. Furthermore, with the help of the novel methodology, the investigation of local sleep phenomena in patients with obstructive sleep apnea could lead to new insights into the pathophysiology of excessive daytime sleepiness.