Header

UZH-Logo

Maintenance Infos

Sleep regulation: sleep restriction, extension, and microsleep detection


Skorucak, Jelena. Sleep regulation: sleep restriction, extension, and microsleep detection. 2017, ETH Zürich, Faculty of Science.

Abstract

Chronic sleep restriction is prevalent in modern societies and is associated with negative physical and mental health outcomes. In adults, a number of factors including social obligations and work demands often lead to insufficient sleep. Chronic sleep restriction not only occurs in adults, but also in adolescents, who are often more severely sleep restricted than adults due to early school start times coupled with academic and social obligations. Thus, given that sleep restriction is a common and may have negative health consequences, it is crucial to understand sleep regulation under conditions of chronic restriction in humans.
The term “sleep homeostasis” refers to the sleep-wake dependent aspect of sleep regulation. Mechanisms of sleep homeostasis regulate sleep intensity: curtailment of sleep results in a rebound of sleep quantity and intensity. Sleep intensity can be quantified by calculating power in slow frequencies of the EEG (slow-wave activity). Our knowledge of this sleep homeostatic process is primarily based on studies using total sleep deprivation. These studies use slow-wave activity to mathematically model the build-up and dissipation of sleep pressure in humans. In contrast to studies of total sleep deprivation, the effects of chronic sleep restriction are less well studied. Several animal studies suggest that adaptation may occur under conditions of chronic sleep restriction, whereby an attenuated or even no homeostatic response (i.e., buildup of slow-wave activity) is observed. This phenomenon is referred to as allostasis and it is still debated whether allostasis is a component of sleep regulation. Some studies suggest that both homeostatic and allostatic responses were observed in response to chronic sleep restriction in rodents, while other rodent studies showed that sleep homeostasis was preserved.
Therefore, the animal literature suggests an allostatic component under conditions of chronic sleep restriction, however, whether allostasis is also present in humans remains unknown. One of the goals of this thesis was to investigate sleep regulation in humans, by analyzing data from an adult chronic sleep restriction and extension study (within subject design, Chapter 3), and an adolescent dose-response study (between subject design, Chapter 4).
Chronic sleep restriction induces daytime sleepiness, and interferes with the daily functioning and performance. Excessive daytime sleepiness may result in the appearance of short sleep fragments during wakefulness, referred to as microsleep episodes. Besides their occurrence in healthy sleep restricted or deprived individuals, microsleep episodes also occur in patients suffering from excessive daytime sleepiness as a consequence of sleep disorders. It is important to be able to detect such episodes for diagnostic purposes and monitoring of the therapy progress, but also for testing patients’ fitness to drive. There are several tests routinely used for the assessment of sleepiness in clinical settings, and all are based on electroencephalography. However, the assessment of sleepiness remains a challenge due to the time consuming and demanding visual scoring of microsleep episodes. Therefore, one of the aims of this thesis was to develop a method for the automatic detection of microsleep episodes based on electroencephalographic and electrooculographic recordings (Chapter 5).
In Chapter 1 of this thesis, basic aspects of sleep research are summarized, discussing the open question of sleep regulation (homeostasis or allostasis), as well as an introduction to microsleep.
Chapter 2 describes the quantitative analysis of sleep, and the methods applied in this thesis, including the well-established spectral analysis and more recent techniques like machine learning approaches.
In Chapter 3, we investigated the response to 7 days of sleep restriction (6 h time in bed) and extension (10 h time in bed) as well as the subsequent response to total sleep deprivation of approximately 40 h in 35 adults (cross-over design). The protocol either restricted or extended sleep by altering participants’ sleep opportunity. Different homeostatic markers were calculated for frontal, central and occipital brain regions: time spent in slow-wave sleep as a percentage of total sleep time, slow-wave activity, rise rate of slow-wave activity, and slow-wave energy. Time spent in slow-wave sleep was increased during sleep restriction and decreased during extension. Generally speaking, there were no significant changes in slow-wave activity, except for a decrease on some nights during the sleep extension protocol. The rise rate of slow-wave activity, a measure of sleep pressure, did not reveal any changes in either the extension or restriction protocols, but was increased after total sleep deprivation. It is possible that this measure is not sensitive enough to moderate changes in sleep pressure but might adequately reflect stronger challenges such as total sleep deprivation. Slow-wave energy, a measure of the total dissipation of sleep pressure across the night, proved to be the most sensitive measure to sleep restriction conditions. It remained at baseline levels during extension, but decreased during restriction, reflecting insufficient dissipation of sleep pressure during restriction. These homeostatic markers were also extracted from simulations of the homeostatic component of the two-process model of sleep regulation and compared with empirical data. There were only minor deviations between the model predictions and the empirical data, without any trend in the temporal evolution of the deviations, which led us to conclude that sleep homeostasis was preserved and there was no indication for an allostatic process.
Chapter 4 is devoted to different doses of sleep restriction in 34 adolescents, distributed into three groups with 5 h, 7.5 h and 10 h of sleep opportunity per night. Each group underwent a protocol of 9 nights designed to mimic a school week between 2 weekends: 2 baseline nights (10 h sleep opportunity), 5 condition nights (5 h, 7.5 h or 10 h), and two recovery nights (10 h). Measures of total sleep time confirmed that the protocol was successful in restricting sleep in the 5-h and 7.5-h conditions by changing participants’ sleep opportunity window, while total sleep time remained at stable levels during the 10-h condition. Markers of sleep homeostasis were determined from frontal and central EEG derivations. Reflecting an increased sleep pressure, time in slow-wave sleep was increased in the 5-h and 7.5-h conditions. Slow-wave activity in the 5-h condition was above baseline levels only in frontal brain regions, although at central derivations a non-significant increase was present. In the 5-h and 7.5-h conditions slow-wave energy was below baseline, indicating an insufficient dissipation of sleep pressure. Slow-wave activity and slow-wave energy also were derived from simulations of the homeostatic process and compared with the empirical data. The model predictions were assessed to be accurate displaying only minor differences from the empirical data, again confirming the presence of a homeostatic processes.
Overall, based on the data from Chapters 3 and 4 we conclude that the homeostatic process remained operative in adults and adolescents in conditions of sleep restriction and extension with hardly any evidence for allostasis. These findings have implications for our understanding of the effects of short and long sleep in both adolescents and adults, and their negative consequences for brain function and health.
In Chapter 5, a method for the automatic detection of microsleep episodes was developed. Data sets from 13 patients contained microsleep episodes marked by clinical experts. Two supervised learning algorithms were applied for the classification, support vector machines and random forests. The classifiers were trained on data of 12 patients and validated in one patient resulting in a specificity of 0.74 and 0.72, respectively, and a sensitivity of 0.99 for both classifiers. This approach is promising for a clinical setting as it provides a semi-automatic procedure, where microsleep episodes would be detected by an algorithm and validated by clinical experts. This would make the process of identification of microsleep episodes much faster and more standardized. However, the main limitation of this approach was the low number of patients for training and testing. For practical applications, validation on a much larger number of patients is needed.
An overall discussion and concluding remarks are presented in Chapter 6.

Abstract

Chronic sleep restriction is prevalent in modern societies and is associated with negative physical and mental health outcomes. In adults, a number of factors including social obligations and work demands often lead to insufficient sleep. Chronic sleep restriction not only occurs in adults, but also in adolescents, who are often more severely sleep restricted than adults due to early school start times coupled with academic and social obligations. Thus, given that sleep restriction is a common and may have negative health consequences, it is crucial to understand sleep regulation under conditions of chronic restriction in humans.
The term “sleep homeostasis” refers to the sleep-wake dependent aspect of sleep regulation. Mechanisms of sleep homeostasis regulate sleep intensity: curtailment of sleep results in a rebound of sleep quantity and intensity. Sleep intensity can be quantified by calculating power in slow frequencies of the EEG (slow-wave activity). Our knowledge of this sleep homeostatic process is primarily based on studies using total sleep deprivation. These studies use slow-wave activity to mathematically model the build-up and dissipation of sleep pressure in humans. In contrast to studies of total sleep deprivation, the effects of chronic sleep restriction are less well studied. Several animal studies suggest that adaptation may occur under conditions of chronic sleep restriction, whereby an attenuated or even no homeostatic response (i.e., buildup of slow-wave activity) is observed. This phenomenon is referred to as allostasis and it is still debated whether allostasis is a component of sleep regulation. Some studies suggest that both homeostatic and allostatic responses were observed in response to chronic sleep restriction in rodents, while other rodent studies showed that sleep homeostasis was preserved.
Therefore, the animal literature suggests an allostatic component under conditions of chronic sleep restriction, however, whether allostasis is also present in humans remains unknown. One of the goals of this thesis was to investigate sleep regulation in humans, by analyzing data from an adult chronic sleep restriction and extension study (within subject design, Chapter 3), and an adolescent dose-response study (between subject design, Chapter 4).
Chronic sleep restriction induces daytime sleepiness, and interferes with the daily functioning and performance. Excessive daytime sleepiness may result in the appearance of short sleep fragments during wakefulness, referred to as microsleep episodes. Besides their occurrence in healthy sleep restricted or deprived individuals, microsleep episodes also occur in patients suffering from excessive daytime sleepiness as a consequence of sleep disorders. It is important to be able to detect such episodes for diagnostic purposes and monitoring of the therapy progress, but also for testing patients’ fitness to drive. There are several tests routinely used for the assessment of sleepiness in clinical settings, and all are based on electroencephalography. However, the assessment of sleepiness remains a challenge due to the time consuming and demanding visual scoring of microsleep episodes. Therefore, one of the aims of this thesis was to develop a method for the automatic detection of microsleep episodes based on electroencephalographic and electrooculographic recordings (Chapter 5).
In Chapter 1 of this thesis, basic aspects of sleep research are summarized, discussing the open question of sleep regulation (homeostasis or allostasis), as well as an introduction to microsleep.
Chapter 2 describes the quantitative analysis of sleep, and the methods applied in this thesis, including the well-established spectral analysis and more recent techniques like machine learning approaches.
In Chapter 3, we investigated the response to 7 days of sleep restriction (6 h time in bed) and extension (10 h time in bed) as well as the subsequent response to total sleep deprivation of approximately 40 h in 35 adults (cross-over design). The protocol either restricted or extended sleep by altering participants’ sleep opportunity. Different homeostatic markers were calculated for frontal, central and occipital brain regions: time spent in slow-wave sleep as a percentage of total sleep time, slow-wave activity, rise rate of slow-wave activity, and slow-wave energy. Time spent in slow-wave sleep was increased during sleep restriction and decreased during extension. Generally speaking, there were no significant changes in slow-wave activity, except for a decrease on some nights during the sleep extension protocol. The rise rate of slow-wave activity, a measure of sleep pressure, did not reveal any changes in either the extension or restriction protocols, but was increased after total sleep deprivation. It is possible that this measure is not sensitive enough to moderate changes in sleep pressure but might adequately reflect stronger challenges such as total sleep deprivation. Slow-wave energy, a measure of the total dissipation of sleep pressure across the night, proved to be the most sensitive measure to sleep restriction conditions. It remained at baseline levels during extension, but decreased during restriction, reflecting insufficient dissipation of sleep pressure during restriction. These homeostatic markers were also extracted from simulations of the homeostatic component of the two-process model of sleep regulation and compared with empirical data. There were only minor deviations between the model predictions and the empirical data, without any trend in the temporal evolution of the deviations, which led us to conclude that sleep homeostasis was preserved and there was no indication for an allostatic process.
Chapter 4 is devoted to different doses of sleep restriction in 34 adolescents, distributed into three groups with 5 h, 7.5 h and 10 h of sleep opportunity per night. Each group underwent a protocol of 9 nights designed to mimic a school week between 2 weekends: 2 baseline nights (10 h sleep opportunity), 5 condition nights (5 h, 7.5 h or 10 h), and two recovery nights (10 h). Measures of total sleep time confirmed that the protocol was successful in restricting sleep in the 5-h and 7.5-h conditions by changing participants’ sleep opportunity window, while total sleep time remained at stable levels during the 10-h condition. Markers of sleep homeostasis were determined from frontal and central EEG derivations. Reflecting an increased sleep pressure, time in slow-wave sleep was increased in the 5-h and 7.5-h conditions. Slow-wave activity in the 5-h condition was above baseline levels only in frontal brain regions, although at central derivations a non-significant increase was present. In the 5-h and 7.5-h conditions slow-wave energy was below baseline, indicating an insufficient dissipation of sleep pressure. Slow-wave activity and slow-wave energy also were derived from simulations of the homeostatic process and compared with the empirical data. The model predictions were assessed to be accurate displaying only minor differences from the empirical data, again confirming the presence of a homeostatic processes.
Overall, based on the data from Chapters 3 and 4 we conclude that the homeostatic process remained operative in adults and adolescents in conditions of sleep restriction and extension with hardly any evidence for allostasis. These findings have implications for our understanding of the effects of short and long sleep in both adolescents and adults, and their negative consequences for brain function and health.
In Chapter 5, a method for the automatic detection of microsleep episodes was developed. Data sets from 13 patients contained microsleep episodes marked by clinical experts. Two supervised learning algorithms were applied for the classification, support vector machines and random forests. The classifiers were trained on data of 12 patients and validated in one patient resulting in a specificity of 0.74 and 0.72, respectively, and a sensitivity of 0.99 for both classifiers. This approach is promising for a clinical setting as it provides a semi-automatic procedure, where microsleep episodes would be detected by an algorithm and validated by clinical experts. This would make the process of identification of microsleep episodes much faster and more standardized. However, the main limitation of this approach was the low number of patients for training and testing. For practical applications, validation on a much larger number of patients is needed.
An overall discussion and concluding remarks are presented in Chapter 6.

Statistics

Additional indexing

Item Type:Dissertation
Referees:Rudin Markus, Achermann Peter, Dijk D J
Communities & Collections:04 Faculty of Medicine > Institute of Pharmacology and Toxicology
07 Faculty of Science > Institute of Pharmacology and Toxicology
Dewey Decimal Classification:570 Life sciences; biology
610 Medicine & health
Language:English
Date:11 December 2017
Deposited On:14 Feb 2018 14:48
Last Modified:19 Mar 2018 10:40
OA Status:Closed

Download

Full text not available from this repository.
Get full-text in a library