From Pulses to Sleep Stages: Towards Optimized Sleep Classification Using Heart-Rate Variability

More and more people quantify their sleep using wearables and are becoming obsessed in their pursuit of optimal sleep ("orthosomnia"). However, it is criticized that many of these wearables are giving inaccurate feedback and can even lead to negative daytime consequences. Acknowledging these facts, we here optimize our previously suggested sleep classification procedure in a new sample of 136 self-reported poor sleepers to minimize erroneous classification during ambulatory sleep sensing. Firstly, we introduce an advanced interbeat-interval (IBI) quality control using a random forest method to account for wearable recordings in naturalistic and more noisy settings. We further aim to improve sleep classification by opting for a loss function model instead of the overall epoch-by-epoch accuracy to avoid model biases towards the majority class (i.e., "light sleep"). Using these implementations, we compare the classification performance between the optimized (loss function model) and the accuracy model. We use signals derived from PSG, one-channel ECG, and two consumer wearables: the ECG breast belt Polar® H10 (H10) and the Polar® Verity Sense (VS), an optical Photoplethysmography (PPG) heart-rate sensor. The results reveal a high overall accuracy for the loss function in ECG (86.3 %, κ = 0.79), as well as the H10 (84.4%, κ = 0.76), and VS (84.2%, κ = 0.75) sensors, with improvements in deep sleep and wake. In addition, the new optimized model displays moderate to high correlations and agreement with PSG on primary sleep parameters, while measures of reliability, expressed in intra-class correlations, suggest excellent reliability for most sleep parameters. Finally, it is demonstrated that the new model is still classifying sleep accurately in 4-classes in users taking heart-affecting and/or psychoactive medication, which can be considered a prerequisite in older individuals with or without common disorders. Further improving and validating automatic sleep stage classification algorithms based on signals from affordable wearables may resolve existing scepticism and open the door for such approaches in clinical practice.

Soccer, Sleep, Repeat: Effects of Training Characteristics on Sleep Quantity and Sleep Architecture

Due to the high demands of competitive sports, the sleep architecture of adolescent athletes may be influenced by their regular training. To date, there is no clear evidence on how training characteristics (intensity, time of day, number of sessions) influence sleep quality and quantity. 53 male soccer players (M = 14.36 years, SD = 0.55) of Austrian U15 (n = 45) and U16 elite teams (n = 8) were tested on at least three consecutive days following their habitual training schedules. Participants completed daily sleep protocols (7 a.m., 8 p.m.) and questionnaires assessing sleep quality (PSQI), chronotype (D-MEQ), competition anxiety (WAI-T), and stress/recovery (RESTQ). Electrocardiography (ECG) and actigraphy devices measured sleep. Using sleep protocols and an ECG-based multi-resolution convolutional neural network (MCNN), we found that higher training intensity leads to more wake time, that later training causes longer sleep duration, and that one training session per day was most advantageous for sleep quality. In addition, somatic complaints assessed by the WAI-T negatively affected adolescent athletes' sleep. Individual training loads and longer recovery times after late training sessions during the day should be considered in training schedules, especially for adolescent athletes. MCNN modeling based on ECG data seems promising for efficient sleep analysis in athletes.

The Virtual Sleep Lab-A Novel Method for Accurate Four-Class Sleep Staging Using Heart-Rate Variability from Low-Cost Wearables

Sleep staging based on polysomnography (PSG) performed by human experts is the de facto "gold standard" for the objective measurement of sleep. PSG and manual sleep staging is, however, personnel-intensive and time-consuming and it is thus impractical to monitor a person's sleep architecture over extended periods. Here, we present a novel, low-cost, automatized, deep learning alternative to PSG sleep staging that provides a reliable epoch-by-epoch four-class sleep staging approach (Wake, Light [N1 + N2], Deep, REM) based solely on inter-beat-interval (IBI) data. Having trained a multi-resolution convolutional neural network (MCNN) on the IBIs of 8898 full-night manually sleep-staged recordings, we tested the MCNN on sleep classification using the IBIs of two low-cost (<EUR 100) consumer wearables: an optical heart rate sensor (VS) and a breast belt (H10), both produced by POLAR®. The overall classification accuracy reached levels comparable to expert inter-rater reliability for both devices (VS: 81%, κ = 0.69; H10: 80.3%, κ = 0.69). In addition, we used the H10 and recorded daily ECG data from 49 participants with sleep complaints over the course of a digital CBT-I-based sleep training program implemented in the App NUKKUAA™. As proof of principle, we classified the IBIs extracted from H10 using the MCNN over the course of the training program and captured sleep-related changes. At the end of the program, participants reported significant improvements in subjective sleep quality and sleep onset latency. Similarly, objective sleep onset latency showed a trend toward improvement. Weekly sleep onset latency, wake time during sleep, and total sleep time also correlated significantly with the subjective reports. The combination of state-of-the-art machine learning with suitable wearables allows continuous and accurate monitoring of sleep in naturalistic settings with profound implications for answering basic and clinical research questions.

Sleep in patients with disorders of consciousness characterized by means of machine learning

Sleep has been proposed to indicate preserved residual brain functioning in patients suffering from disorders of consciousness (DOC) after awakening from coma. However, a reliable characterization of sleep patterns in this clinical population continues to be challenging given severely altered brain oscillations, frequent and extended artifacts in clinical recordings and the absence of established staging criteria. In the present study, we try to address these issues and investigate the usefulness of a multivariate machine learning technique based on permutation entropy, a complexity measure. Specifically, we used long-term polysomnography (PSG), along with video recordings in day and night periods in a sample of 23 DOC; 12 patients were diagnosed as Unresponsive Wakefulness Syndrome (UWS) and 11 were diagnosed as Minimally Conscious State (MCS). Eight hour PSG recordings of healthy sleepers (N = 26) were additionally used for training and setting parameters of supervised and unsupervised model, respectively. In DOC, the supervised classification (wake, N1, N2, N3 or REM) was validated using simultaneous videos which identified periods with prolonged eye opening or eye closure.The supervised classification revealed that out of the 23 subjects, 11 patients (5 MCS and 6 UWS) yielded highly accurate classification with an average F1-score of 0.87 representing high overlap between the classifier predicting sleep (i.e. one of the 4 sleep stages) and closed eyes. Furthermore, the unsupervised approach revealed a more complex pattern of sleep-wake stages during the night period in the MCS group, as evidenced by the presence of several distinct clusters. In contrast, in UWS patients no such clustering was found. Altogether, we present a novel data-driven method, based on machine learning that can be used to gain new and unambiguous insights into sleep organization and residual brain functioning of patients with DOC.

The effect of daytime napping and full-night sleep on the consolidation of declarative and procedural information

Many studies investigating sleep and memory consolidation have evaluated full-night sleep rather than alternative sleep periods such as daytime naps. This multi-centre study followed up on, and was compared with, an earlier full-night study (Schabus et al., 2004) investigating the relevance of daytime naps for the consolidation of declarative and procedural memory. Seventy-six participants were randomly assigned to a nap or wake group, and performed a declarative word-pair association or procedural mirror-tracing task. Performance changes from before to after a 90-min retention interval filled with sleep or quiet wakefulness were evaluated between groups. Associations between performance changes, sleep architecture, spindles, and slow oscillations were investigated. For the declarative task we observed a trend towards stronger forgetting across a wake period compared with a nap period, and a trend towards memory increase over the full-night. For the procedural task, accuracy was significantly decreased following daytime wakefulness, showed a trend to increase with a daytime nap, and significantly increased across full-night sleep. For the nap protocol, neither sleep stages, spindles, nor slow oscillations predicted performance changes. A direct comparison of day and nighttime sleep revealed that daytime naps are characterized by significantly lower spindle density, but higher spindle activity and amplitude compared with full-night sleep. In summary, data indicate that daytime naps protect procedural memories from deterioration, whereas full-night sleep improves performance. Given behavioural and physiological differences between day and nighttime sleep, future studies should try to characterize potential differential effects of full-night and daytime sleep with regard to sleep-dependent memory consolidation.

Preferential processing of emotionally and self-relevant stimuli persists in unconscious N2 sleep

Information processing has been suggested to depend on the current state of the brain as well as stimulus characteristics (e.g. salience). We compared processing of salient stimuli (subject's own names [SONs] and angry voice [AV] stimuli) to processing of unfamiliar names (UNs) and neutral voice (NV) stimuli across different vigilance stages (i.e. wakefulness as well as sleep stages N1 and N2) by means of event-related oscillatory responses during wakefulness and a subsequent afternoon nap. Our findings suggest that emotional prosody and self-relevance drew more attentional resources during wakefulness with specifically AV stimuli being processed more strongly. During N1, SONs were more arousing than UNs irrespective of prosody. Moreover, emotional and self-relevant stimuli evoked stronger responses also during N2 sleep suggesting a 'sentinel processing mode' of the brain during this state of naturally occurring unconsciousness. Finally, this initial preferential processing of salient stimuli during N2 sleep seems to be followed by an inhibitory sleep-protecting process, which is reflected by a K-complex-like response.

Better than sham? A double-blind placebo-controlled neurofeedback study in primary insomnia

See Thibault et al. (doi:10.1093/awx033) for a scientific commentary on this article.Neurofeedback training builds upon the simple concept of instrumental conditioning, i.e. behaviour that is rewarded is more likely to reoccur, an effect Thorndike referred to as the 'law of effect'. In the case of neurofeedback, information about specific electroencephalographic activity is fed back to the participant who is rewarded whenever the desired electroencephalography pattern is generated. If some kind of hyperarousal needs to be addressed, the neurofeedback community considers sensorimotor rhythm neurofeedback as the gold standard. Earlier treatment approaches using sensorimotor-rhythm neurofeedback indicated that training to increase 12-15 Hz sensorimotor rhythm over the sensorimotor cortex during wakefulness could reduce attention-deficit/hyperactivity disorder and epilepsy symptoms and even improve sleep quality by enhancing sleep spindle activity (lying in the same frequency range). In the present study we sought to critically test whether earlier findings on the positive effect of sensorimotor rhythm neurofeedback on sleep quality and memory could also be replicated in a double-blind placebo-controlled study on 25 patients with insomnia. Patients spent nine polysomnography nights and 12 sessions of neurofeedback and 12 sessions of placebo-feedback training (sham) in our laboratory. Crucially, we found both neurofeedback and placebo feedback to be equally effective as reflected in subjective measures of sleep complaints suggesting that the observed improvements were due to unspecific factors such as experiencing trust and receiving care and empathy from experimenters. In addition, these improvements were not reflected in objective electroencephalographic-derived measures of sleep quality. Furthermore, objective electroencephalographic measures that potentially reflected mechanisms underlying the efficacy of neurofeedback such as spectral electroencephalographic measures and sleep spindle parameters remained unchanged following 12 training sessions. A stratification into 'true' insomnia patients and 'insomnia misperceivers' (subjective, but no objective sleep problems) did not alter the results. Based on this comprehensive and well-controlled study, we conclude that for the treatment of primary insomnia, neurofeedback does not have a specific efficacy beyond unspecific placebo effects. Importantly, we do not find an advantage of neurofeedback over placebo feedback, therefore it cannot be recommended as an alternative to cognitive behavioural therapy for insomnia, the current (non-pharmacological) standard-of-care treatment. In addition, our study may foster a critical discussion that generally questions the effectiveness of neurofeedback, and emphasizes the importance of demonstrating neurofeedback efficacy in other study samples and disorders using truly placebo and double-blind controlled trials.

Night and day variations of sleep in patients with disorders of consciousness

Brain injuries substantially change the entire landscape of oscillatory dynamics and render detection of typical sleep patterns difficult. Yet, sleep is characterized not only by specific EEG waveforms, but also by its circadian organization. In the present study we investigated whether brain dynamics of patients with disorders of consciousness systematically change between day and night. We recorded ~24 h EEG at the bedside of 18 patients diagnosed to be vigilant but unaware (Unresponsive Wakefulness Syndrome) and 17 patients revealing signs of fluctuating consciousness (Minimally Conscious State). The day-to-night changes in (i) spectral power, (ii) sleep-specific oscillatory patterns and (iii) signal complexity were analyzed and compared to 26 healthy control subjects. Surprisingly, the prevalence of sleep spindles and slow waves did not systematically vary between day and night in patients, whereas day-night changes in EEG power spectra and signal complexity were revealed in minimally conscious but not unaware patients.

The Voice of Anger: Oscillatory EEG Responses to Emotional Prosody

Emotionally relevant stimuli and in particular anger are, due to their evolutionary relevance, often processed automatically and able to modulate attention independent of conscious access. Here, we tested whether attention allocation is enhanced when auditory stimuli are uttered by an angry voice. We recorded EEG and presented healthy individuals with a passive condition where unfamiliar names as well as the subject's own name were spoken both with an angry and neutral prosody. The active condition instead, required participants to actively count one of the presented (angry) names. Results revealed that in the passive condition the angry prosody only elicited slightly stronger delta synchronization as compared to a neutral voice. In the active condition the attended (angry) target was related to enhanced delta/theta synchronization as well as alpha desynchronization suggesting enhanced allocation of attention and utilization of working memory resources. Altogether, the current results are in line with previous findings and highlight that attention orientation can be systematically related to specific oscillatory brain responses. Potential applications include assessment of non-communicative clinical groups such as post-comatose patients.

EEG oscillations reflect the complexity of social interactions in a non-verbal social cognition task using animated triangles

The ability to attribute independent mental states (e.g. opinions, perceptions, beliefs) to oneself and others is termed Theory of Mind (ToM). Previous studies investigating ToM usually employed verbal paradigms and functional neuroimaging methods. Here, we studied oscillatory responses in the electroencephalogram (EEG) in a non-verbal social cognition task. The aim of this study was twofold: First, we wanted to investigate differences in oscillatory responses to animations differing with regard to the complexity of social "interactions". Secondly, we intended to evaluate the basic cognitive processes underlying social cognition. To this end, we analyzed theta, alpha, beta and gamma task-related de-/synchronization (TRD/TRS) during presentation of six non-verbal videos differing in the complexity of (social) "interactions" between two geometric shapes. Videos were adopted from Castelli et al. (2000)and belonged to three conditions: Videos designed to evoke attributions of mental states (ToM), interaction descriptions (goal-directed, GD) and videos in which the shapes moved randomly (R). Analyses revealed that only theta activity consistently varied as a function of social "interaction" complexity. Results suggest that ToM/GD videos attract more attention and working-memory resources and may have activated related memory contents. Alpha and beta results were less consistent. While alpha effects suggest that observation of social "interactions" may benefit from inhibition of self-centered processing, oscillatory responses in the beta range could be related to action observation. In summary, the results provide insight into basic cognitive processes involved in social cognition and render the paradigm attractive for the investigation of social cognitive processes in non-verbal populations.

Oscillatory theta activity during memory formation and its impact on overnight consolidation: a missing link?

Sleep has been shown to promote memory consolidation driven by certain oscillatory patterns, such as sleep spindles. However, sleep does not consolidate all newly encoded information uniformly but rather "selects" certain memories for consolidation. It is assumed that such selection depends on salience tags attached to the new memories before sleep. However, little is known about the underlying neuronal processes reflecting presleep memory tagging. The current study sought to address the question of whether event-related changes in spectral theta power (theta ERSP) during presleep memory formation could reflect memory tagging that influences subsequent consolidation during sleep. Twenty-four participants memorized 160 word pairs before sleep; in a separate laboratory visit, they performed a nonlearning control task. Memory performance was tested twice, directly before and after 8 hr of sleep. Results indicate that participants who improved their memory performance overnight displayed stronger theta ERSP during the memory task in comparison with the control task. They also displayed stronger memory task-related increases in fast sleep spindle activity. Furthermore, presleep theta activity was directly linked to fast sleep spindle activity, indicating that processes during memory formation might indeed reflect memory tagging that influences subsequent consolidation during sleep. Interestingly, our results further indicate that the suggested relation between sleep spindles and overnight performance change is not as direct as once believed. Rather, it appears to be mediated by processes beginning during presleep memory formation. We conclude that theta ERSP during presleep memory formation reflects cortico-hippocampal interactions that lead to a better long-term accessibility by tagging memories for sleep spindle-related reprocessing.

Oscillatory brain responses to own names uttered by unfamiliar and familiar voices

Among auditory stimuli, the own name is one of the most powerful and it is able to automatically capture attention and elicit a robust electrophysiological response. The subject’s own name (SON) is preferentially processed in the right hemisphere, mainly because of its self-relevance and emotional content, together with other personally relevant information such as the voice of a familiar person. Whether emotional and self-relevant information are able to attract attention and can be, in future, introduced in clinical studies remains unclear. In the present study we used EEG and asked participants to count a target name (active condition) or to just listen to the SON or other unfamiliar names uttered by a familiar or unfamiliar voice (passive condition). Data reveals that the target name elicits a strong alpha event related desynchronization with respect to non-target names and triggers in addition a left lateralized theta synchronization as well as delta synchronization.

In the passive condition alpha desynchronization was observed for familiar voice and SON stimuli in the right hemisphere.

Altogether we speculate that participants engage additional attentional resources when counting a target name or when listening to personally relevant stimuli which is indexed by alpha desynchronization whereas left lateralized theta synchronization may be related to verbal working memory load. After validating the present protocol in healthy volunteers it is suggested to move one step further and apply the protocol to patients with disorders of consciousness in which the degree of residual cognitive processing and self-awareness is still insufficiently understood.

•

EEG during an active–passive task based on first names was time-frequency analyzed.

•

The presented names were uttered either by an unfamiliar or a familiar voice.

•

Counted names elicited alpha desynchronization and left theta synchronization.

•

Own name and familiar voices enhanced strong right alpha desynchronization.

•

Alpha desynchronization reflects attentional engagement and emotional processing.

Slow sleep spindle activity, declarative memory, and general cognitive abilities in children

STUDY OBJECTIVES

Functional interactions between sleep spindle activity, declarative memory consolidation, and general cognitive abilities in school-aged children.

DESIGN

Healthy, prepubertal children (n = 63; mean age 9.56 ± 0.76 y); ambulatory all-night polysomnography (2 nights); investigating the effect of prior learning (word pair association task; experimental night) versus nonlearning (baseline night) on sleep spindle activity; general cognitive abilities assessed using the Wechsler Intelligence Scale for Children-IV (WISC-IV).

MEASUREMENTS AND RESULTS

Analysis of spindle activity during nonrapid eye movement sleep (N2 and N3) evidenced predominant peaks in the slow (11-13 Hz) but not in the fast (13-15 Hz) sleep spindle frequency range (baseline and experimental night). Analyses were restricted to slow sleep spindles. Changes in spindle activity from the baseline to the experimental night were not associated with the overnight change in the number of recalled words reflecting declarative memory consolidation. Children with higher sleep spindle activity as measured at frontal, central, parietal, and occipital sites during both baseline and experimental nights exhibited higher general cognitive abilities (WISC-IV) and declarative learning efficiency (i.e., number of recalled words before and after sleep).

CONCLUSIONS

Slow sleep spindles (11-13 Hz) in children age 8-11 y are associated with inter-individual differences in general cognitive abilities and learning efficiency.

Susceptibility to declarative memory interference is pronounced in primary insomnia

Sleep has been shown to stabilize memory traces and to protect against competing interference in both the procedural and declarative memory domain. Here, we focused on an interference learning paradigm by testing patients with primary insomnia (N = 27) and healthy control subjects (N = 21). In two separate experimental nights with full polysomnography it was revealed that after morning interference procedural memory performance (using a finger tapping task) was not impaired in insomnia patients while declarative memory (word pair association) was decreased following interference. More specifically, we demonstrate robust associations of central sleep spindles (in N3) with motor memory susceptibility to interference as well as (cortically more widespread) fast spindle associations with declarative memory susceptibility. In general the results suggest that insufficient sleep quality does not necessarily show up in worse overnight consolidation in insomnia but may only become evident (in the declarative memory domain) when interference is imposed.

Consolidation of temporal order in episodic memories

Even though it is known that sleep benefits declarative memory consolidation, the role of sleep in the storage of temporal sequences has rarely been examined. Thus we explored the influence of sleep on temporal order in an episodic memory task followed by sleep or sleep deprivation. Thirty-four healthy subjects (17 men) aged between 19 and 28 years participated in the randomized, counterbalanced, between-subject design. Parameters of interests were NREM/REM cycles, spindle activity and spindle-related EEG power spectra. Participants of both groups (sleep group/sleep deprivation group) performed retrieval in the evening, morning and three days after the learning night. Results revealed that performance in temporal order memory significantly deteriorated over three days only in sleep deprived participants. Furthermore our data showed a positive relationship between the ratios of the (i) first NREM/REM cycle with more REM being associated with delayed temporal order recall. Most interestingly, data additionally indicated that (ii) memory enhancers in the sleep group show more fast spindle related alpha power at frontal electrode sites possibly indicating access to a yet to be consolidated memory trace. We suggest that distinct sleep mechanisms subserve different aspects of episodic memory and are jointly involved in sleep-dependent memory consolidation.