K-band Doppler radar for contact-less overnight sleep marker assessment: a pilot validation study.

An estimated 45 million persons in Europe are annually subjected to sleep-wake disorders. State-of-the-art polysomnography provides sophisticated insights into sleep (patho)physiology. A drawback of the method, however, is the obtrusive setting dependent on a clinical-based sleep laboratory with high operational costs. A contact-less prototype was developed to monitor limb movements and vital signs during sleep. A dual channel K-band Doppler radar transceiver captured limb movements and periodic chest wall motion due to respiration and heart activity. A wavelet transform based multi-resolution analysis (MRA) approach isolated limb movements, respiration, and heart rate from the demodulated signal. A test bench setup characterized the prototype simulating near physiological chest wall motions caused by periodic respiration and heartbeats in humans. Single- and multi-tone test bench simulations showed extremely low relative percentage errors of the prototype for respiratory and heart rate within -2 and 1%. The performance of the prototype was validated in overnight comparative studies, involving two healthy volunteers, with polysomnography as the reference. The prototype has successfully classified limb movements, with a sensitivity and specificity of 88.9 and 76.8% respectively, and has achieved accurate respiratory and heart rate measurement performance with overall absolute errors of 1 breath per minute for respiration and 3 beats per minute for heart rate. This pilot study shows that K-band Doppler radar and wavelet transform MRA seem to be valid for overnight sleep marker assessment. The contact-less approach might offer a promising solution for home-based sleep monitoring and assessment.

Epileptic seizures as condensed sleep: an analysis of network dynamics from electroencephalogram signals.

Both deepening sleep and evolving epileptic seizures are associated with increasing slow-wave activity. Larger-scale functional networks derived from electroencephalogram indicate that in both transitions dramatic changes of communication between brain areas occur. During seizures these changes seem to be 'condensed', because they evolve more rapidly than during deepening sleep. Here we set out to assess quantitatively functional network dynamics derived from electroencephalogram signals during seizures and normal sleep. Functional networks were derived from electroencephalogram signals from wakefulness, light and deep sleep of 12 volunteers, and from pre-seizure, seizure and post-seizure time periods of 10 patients suffering from focal onset pharmaco-resistant epilepsy. Nodes of the functional network represented electrical signals recorded by single electrodes and were linked if there was non-random cross-correlation between the two corresponding electroencephalogram signals. Network dynamics were then characterized by the evolution of global efficiency, which measures ease of information transmission. Global efficiency was compared with relative delta power. Global efficiency significantly decreased both between light and deep sleep, and between pre-seizure, seizure and post-seizure time periods. The decrease of global efficiency was due to a loss of functional links. While global efficiency decreased significantly, relative delta power increased except between the time periods wakefulness and light sleep, and pre-seizure and seizure. Our results demonstrate that both epileptic seizures and deepening sleep are characterized by dramatic fragmentation of larger-scale functional networks, and further support the similarities between sleep and seizures.

Selective REM sleep deprivation in narcolepsy.

Narcolepsy is characterized by excessive daytime sleepiness and rapid eye movement (REM) sleep abnormalities, including cataplexy. The aim of this study was to assess REM sleep pressure and homeostasis in narcolepsy. Six patients with narcolepsy and six healthy controls underwent a REM sleep deprivation protocol, including one habituation, one baseline, two deprivation nights (D1, D2) and one recovery night. Multiple sleep latency tests (MSLTs) were performed during the day after baseline and after D2. During D1 and D2 REM sleep was prevented by awakening the subjects at the first polysomnographic signs of REM sleep for 2 min. Mean sleep latency and number of sleep-onset REM periods (SOREMs) were determined on all MSLT. More interventions were required to prevent REM sleep in narcoleptics compared with control subjects during D1 (57 ± 16 versus 24 ± 10) and D2 (87 ± 22 versus 35 ± 8, P = 0.004). Interventions increased from D1 to D2 by 46% in controls and by 53% in narcoleptics (P < 0.03). Selective REM sleep deprivation was successful in both controls (mean reduction of REM to 6% of baseline) and narcoleptics (11%). Both groups had a reduction of total sleep time during the deprivation nights (P = 0.03). Neither group had REM sleep rebound in the recovery night. Narcoleptics had, however, an increase in the number of SOREMs on MSLT (P = 0.005). There was no increase in the number of cataplexies after selective REM sleep deprivation. We conclude that: (i) REM sleep pressure is higher in narcoleptics; (ii) REM sleep homeostasis is similar in narcoleptics and controls; (iii) in narcoleptics selective REM sleep deprivation may have an effect on sleep propensity but not on cataplexy.

Personalized Sleep Spindle Detection in Whole Night Polysomnography.

The present study proposes a new personalized sleep spindle detection algorithm, suggesting the importance of an individualized approach. We identify an optimal set of features that characterize the spindle and exploit a support vector machine to distinguish between spindle and nonspindle patterns. The algorithm is assessed on the open source DREAMS database, that contains only selected part of the polysomnography, and on whole night polysomnography recordings from the SPASH database. We show that on the former database the personalization can boost sensitivity, from 84.2% to 89.8%, with a slight increase in specificity, from 97.6% to 98.1%. On a whole night polysomnography instead, the algorithm reaches a sensitivity of 98.6% and a specificity of 98.1%, thanks to the personalization approach. Future work will address the integration of the spindle detection algorithm within a sleep scoring automated procedure.

Pulse Oximetry at the Wrist During Sleep: Performance, Challenges and Perspectives.

The state-of-the-art non-invasive measurement of peripheral oxygen saturation (SpO2) during sleep is mainly based on pulse oximetry at the fingertip. Although this approach is noninvasive, it can still be obtrusive and cumbersome to apply, in particular for ambulatory monitoring over several nights.We developed a wrist-worn reflectance pulse oximetry device which can be embedded in a watch, making it less obtrusive and easy to apply. This device was tested in an ongoing clinical study on 57 subjects (33 patients and 24 healthy volunteers) undergoing a full overnight polysomnography recording. The accuracy was evaluated against state-of-the-art fingertip SpO2 measurements.In the 54 subjects available for analysis we obtained an SpO2 accuracy (ARMS) of 3.4 % when automatically rejecting 17.7 % of signals due to low quality. When further excluding measurements suffering from insufficient contact of the watch with the skin an ARMS of 2.7 % was obtained while rejecting a total of 23.2 % measurements. These accuracies comply with the ISO standard and the FDA guidance for pulse oximeters.The present results are promising and pave the way for unobtrusive and continuous monitoring of SpO2 to screen for sleep disordered breathing. Nonetheless, contact pressure and venous blood have shown to adversely affect the SpO2 estimation and remain a challenge for wrist-based reflectance pulse oximetry.

Automated sleep scoring: A review of the latest approaches.

Clinical sleep scoring involves a tedious visual review of overnight polysomnograms by a human expert, according to official standards. It could appear then a suitable task for modern artificial intelligence algorithms. Indeed, machine learning algorithms have been applied to sleep scoring for many years. As a result, several software products offer nowadays automated or semi-automated scoring services. However, the vast majority of the sleep physicians do not use them. Very recently, thanks to the increased computational power, deep learning has also been employed with promising results. Machine learning algorithms can undoubtedly reach a high accuracy in specific situations, but there are many difficulties in their introduction in the daily routine. In this review, the latest approaches that are applying deep learning for facilitating and accelerating sleep scoring are thoroughly analyzed and compared with the state of the art methods. Then the obstacles in introducing automated sleep scoring in the clinical practice are examined. Deep learning algorithm capabilities of learning from a highly heterogeneous dataset, in terms both of human data and of scorers, are very promising and should be further investigated.

Stroke causes a transient imbalance of interhemispheric information flow in EEG during non-REM sleep.

OBJECTIVE: Large-scale connectivity, especially interhemispheric connections, plays a crucial role for recovery after stroke. Here we used methods from information theory to characterize interhemispheric information flow in wake- and sleep-EEG after cerebral ischemia. METHODS: 34 patients with unilateral ischemic stroke were included. Symbolic Transfer Entropy (STE) was applied between bipolar EEG signals on the left and the right cerebral hemisphere during polysomnographic recordings in the acute phase and 3 months after stroke. RESULTS: In the acute phase, we found a sleep stage-dependent preferred interhemispheric asymmetry: during non-REM sleep the information flow was predominantly directed from the contralesional toward the ipsilesional hemisphere. This effect was greatly reduced in a follow-up recording 3 months after stroke onset. CONCLUSION: Our findings are consistent with functional imaging studies showing a transient hyperactivity of contralesional areas after stroke. We conclude that STE is a robust method for detecting post-stroke connectivity reorganizations, and that sleep stages have to be taken into account when assessing functional connectivity. SIGNIFICANCE: EEG is more widely available than functional MRI. Future studies will have to confirm whether EEG derived STE can be useful in a clinical setting during rehabilitation after stroke.

Effects of sleep fragmentation on the arousability to resistive loading in NREM and REM sleep in normal men.

STUDY OBJECTIVE: In healthy subjects, arousability to inspiratory resistive loading is greater during rapid eye movement (REM) sleep compared with non-REM (NREM) sleep but is poorest in REM sleep in patients with sleep apnea. We therefore examined the hypothesis that sleep fragmentation impairs arousability, especially from REM sleep. DESIGN: Two blocks of 3 polysomnographies (separated by at least 1 week) were performed randomly. An inspiratory-loaded night followed either 2 undisturbed control nights (LN(C)) or 2 acoustically fragmented nights (LN(F)) SETTING: Sleep laboratory. PARTICIPANTS: Sixteen healthy men aged 20 to 29 years. INTERVENTIONS: In both loaded nights, an inspiratory resistive load was added via a valved facemask every 2 minutes during sleep and turned off either when arousal occurred or after 2 minutes. MEASUREMENTS AND RESULTS: During LN(F), arousability remained significantly greater in REM sleep (71% aroused within 2 minutes) compared with stage 2 (29%) or stage 3/4 (16%) sleep. After sleep fragmentation, arousability was decreased in stage 2 sleep (LN(F): 29%; LN(C): 38%; p < .05) and low in early REM sleep, increasing across the night (p < .01). In stage 3/4 sleep, neither an attenuation nor a change across the night was seen after sleep fragmentation. CONCLUSIONS: Mild sleep fragmentation is already sufficient to attenuate arousability in stage 2 sleep and to decrease arousability in early, compared with late, REM sleep. This means that sleep fragmentation affects the arousal response to increasing resistance and that the effects are different in stage 2 and REM sleep. The biologic reason for this increase in the arousal response in REM sleep across the night is not clear.

Subjective perception of sleepiness in a driving simulator is different from that in the Maintenance of Wakefulness Test.

OBJECTIVE: To test whether sleep-deprived, healthy subjects who do not always signal spontaneously perceived sleepiness (SPS) before falling asleep during the Maintenance of Wakefulness Test (MWT) would do so in a driving simulator. METHODS: Twenty-four healthy subjects (20-26 years old) underwent a MWT for 40 min and a driving simulator test for 1 h, before and after one night of sleep deprivation. Standard electroencephalography, electrooculography, submental electromyography, and face videography were recorded simultaneously to score wakefulness and sleep. Subjects were instructed to signal SPS as soon as they subjectively felt sleepy and to try to stay awake for as long as possible in every test. They were rewarded for both "appropriate" perception of SPS and staying awake for as long as possible. RESULTS: After sleep deprivation, seven subjects (29%) did not signal SPS before falling asleep in the MWT, but all subjects signalled SPS before falling asleep in the driving simulator (p <0.004). CONCLUSIONS: The previous results of an "inaccurate" SPS in the MWT were confirmed, and a perfect SPS was shown in the driving simulator. It was hypothesised that SPS is more accurate for tasks involving continuous feedback of performance, such as driving, compared to the less active situation of the MWT. Spontaneously perceived sleepiness in the MWT cannot be used to judge sleepiness perception while driving. Further studies are needed to define the accuracy of SPS in working tasks or occupations with minimal or no performance feedback.

Word encoding during sleep is suggested by correlations between word-evoked up-states and post-sleep semantic priming.

To test whether humans can encode words during sleep we played everyday words to men while they were napping and assessed priming from sleep-played words following waking. Words were presented during non-rapid eye movement (NREM) sleep. Priming was assessed using a semantic and a perceptual priming test. These tests measured differences in the processing of words that had been or had not been played during sleep. Synonyms to sleep-played words were the targets in the semantic priming test that tapped the meaning of sleep-played words. All men responded to sleep-played words by producing up-states in their electroencephalogram. Up-states are NREM sleep-specific phases of briefly increased neuronal excitability. The word-evoked up-states might have promoted word processing during sleep. Yet, the mean performance in the priming tests administered following sleep was at chance level, which suggests that participants as a group failed to show priming following sleep. However, performance in the two priming tests was positively correlated to each other and to the magnitude of the word-evoked up-states. Hence, the larger a participant's word-evoked up-states, the larger his perceptual and semantic priming. Those participants who scored high on all variables must have encoded words during sleep. We conclude that some humans are able to encode words during sleep, but more research is needed to pin down the factors that modulate this ability.

Sleep stage II contributes to the consolidation of declarative memories.

Various studies suggest that non-rapid eye movement (NREM) sleep, especially slow-wave sleep (SWS), is vital to the consolidation of declarative memories. However, sleep stage 2 (S2), which is the other NREM sleep stage besides SWS, has gained only little attention. The current study investigated whether S2 during an afternoon nap contributes to the consolidation of declarative memories. Participants learned associations between faces and cities prior to a brief nap. A cued recall test was administered before and following the nap. Spindle, delta and slow oscillation activity was recorded during S2 in the nap following learning and in a control nap. Increases in spindle activity, delta activity, and slow oscillation activity in S2 in the nap following learning compared to the control nap were associated with enhanced retention of face-city associations. Furthermore, spindles tended to occur more frequently during up-states than down-states within slow oscillations during S2 following learning versus S2 of the control nap. These findings suggest that spindles, delta waves, and slow oscillations might promote memory consolidation not only during SWS, as shown earlier, but also during S2.

EEG correlation and power during maintenance of wakefulness test after sleep-deprivation.

OBJECTIVES: To investigate whether there are any objective EEG characteristics that change significantly between specific time periods during maintenance of wakefulness test (MWT) and whether such changes are associated with the ability to appropriately communicate sleepiness. METHODS: After a night of total sleep deprivation, 12 healthy young subjects underwent a MWT whilst being instructed to communicate the experience of subjective sleepiness by pressing a button. EEG analysis consisted of average relative power and correlation between EEG signals. RESULTS: A comparison of the 30 s before microsleep (MS) with 30 s before subjects communicated experience of sleepiness (PB) showed increased ß correlation as well as increased power in the ß band (13-20 Hz) whereas power in the θ (4.5-7.5 Hz) and α (8-12.5 Hz) band was significantly decreased. When subjects later failed to communicate the experience of subjective sleepiness before (micro-)sleep occurred, average relative power and EEG correlation were significantly higher during 30 s following lights off in the δ (1-4 Hz) band and power in the α and ß bands was decreased. CONCLUSIONS: EEG spectral power and correlation change significantly in specific frequency bands between different time periods of MWT. Failure to communicate sleepiness is associated with certain precursors of EEG power and correlation. SIGNIFICANCE: This study demonstrates that there are specific EEG characteristics associated with impending failure to communicate sleepiness.

Sleepiness is not always perceived before falling asleep in healthy, sleep-deprived subjects.

OBJECTIVE: To test whether subjects spontaneously signal sleepiness before falling asleep under monotonous conditions. METHODS: Twenty-eight healthy students were deprived of sleep for one night and then underwent a "maintenance-of-wakefulness test" (MWT) consisting of four 40-min trials. They were told to give a signal as soon as they felt sleepy and to try to stay awake as long as possible. In a first series of tests, the subjects were given no reward (nr); in a second series, monetary rewards (wr) were given both for an accurate perception of sleepiness and for staying awake longer. RESULTS: Seventeen of the 28 subjects (60.7%) did not signal sleepiness before a sleep fragment occurred in at least one of the four MWT trials. Women were more reliably aware of sleepiness than men in the nr trials (p=.02), while the men's performance improved in the wr trials (p<.02), becoming equivalent to the women's performance. CONCLUSIONS: Our results cast doubt on the general assumption that one cannot fall asleep without feeling sleepy first. If similar results can be obtained in monotonous driving or working situations, this will imply that accidents caused by sleepiness or by falling asleep cannot necessarily be attributed to an individual's negligence.

Visual and spectral analysis of sleep EEG in acute hemispheric stroke.

BACKGROUND: Reports on the effects of focal hemispheric damage on sleep EEG are rare and contradictory. PATIENTS AND METHODS: Twenty patients (mean age +/- SD 53 +/- 14 years) with a first acute hemispheric stroke and no sleep apnea were studied. Stroke severity [National Institute of Health Stroke Scale (NIHSS)], volume (diffusion-weighted brain MRI), and short-term outcome (Rankin score) were assessed. Within the first 8 days after stroke onset, 1-3 sleep EEG recordings per patient were performed. Sleep scoring and spectral analysis were based on the central derivation of the healthy hemisphere. Data were compared with those of 10 age-matched and gender-matched hospitalized controls with no brain damage and no sleep apnea. RESULTS: Stroke patients had higher amounts of wakefulness after sleep onset (112 +/- 53 min vs. 60 +/- 38 min, p < 0.05) and a lower sleep efficiency (76 +/- 10% vs. 86 +/- 8%, p < 0.05) than controls. Time spent in slow-wave sleep (SWS) and rapid eye movement (REM) sleep and total sleep time were lower in stroke patients, but differences were not significant. A positive correlation was found between the amount of SWS and stroke volume (r = 0.79). The slow-wave activity (SWA) ratio NREM sleep/wakefulness was lower in patients than in controls (p < 0.05), and correlated with NIHSS (r = -0.47). CONCLUSION: Acute hemispheric stroke is accompanied by alterations of sleep EEG over the healthy hemisphere that correlate with stroke volume and outcome. The increased SWA during wakefulness and SWS over the healthy hemisphere contralaterally to large strokes may reflect neuronal hypometabolism induced transhemispherically (diaschisis).

Evolution of sleep and sleep EEG after hemispheric stroke.

The evolution of subjective sleep and sleep electroencephalogram (EEG) after hemispheric stroke have been rarely studied and the relationship of sleep variables to stroke outcome is essentially unknown. We studied 27 patients with first hemispheric ischaemic stroke and no sleep apnoea in the acute (1-8 days), subacute (9-35 days), and chronic phase (5-24 months) after stroke. Clinical assessment included estimated sleep time per 24 h (EST) and Epworth sleepiness score (ESS) before stroke, as well as EST, ESS and clinical outcome after stroke. Sleep EEG data from stroke patients were compared with data from 11 hospitalized controls and published norms. Changes in EST (>2 h, 38% of patients) and ESS (>3 points, 26%) were frequent but correlated poorly with sleep EEG changes. In the chronic phase no significant differences in sleep EEG between controls and patients were found. High sleep efficiency and low wakefulness after sleep onset in the acute phase were associated with a good long-term outcome. These two sleep EEG variables improved significantly from the acute to the subacute and chronic phase. In conclusion, hemispheric strokes can cause insomnia, hypersomnia or changes in sleep needs but only rarely persisting sleep EEG abnormalities. High sleep EEG continuity in the acute phase of stroke heralds a good clinical outcome.