Estimates Of Total Sleep Time Research Articles

Do weighted blankets improve sleep among children with a history of maltreatment? A randomized controlled crossover trial.

Sleep disruption is prevalent and persistent among children who experience maltreatment/interpersonal trauma. Weighted blankets have gained popularity in recent years as a potential nonpharmacological intervention for improving sleep in various populations, but their efficacy has not been examined among maltreated children. The current study used a randomized, within-subjects, crossover design to examine whether the use of a weighted blanket improves objective and/or subjective indices of sleep among 30 children, ages 6-15 years (mean = 9.7, standard deviation = 2.9) adopted from foster care. Participants used a weighted blanket for 2 weeks and their usual (unweighted) blanket for 2 weeks in a counterbalanced order. Sleep outcomes were measured using actigraphy and subjective sleep diaries. No differences in actigraphy-based or subjective estimates of total sleep time, sleep onset latency, wake after sleep onset, or sleep quality ratings were found based on blanket type. Child age, biological sex, timing of participation (school year vs summer months), and maltreatment/trauma history did not impact outcomes. Although we did not find evidence that weighted blankets improve sleep among children with a history of maltreatment/interpersonal trauma, additional well-controlled studies using larger samples of children are needed. Cifre AB, Vieira A, Baker C, etal. Do weighted blankets improve sleep among children with a history of maltreatment? A randomized controlled crossover trial. J Clin Sleep Med. 2024;20(9):1405-1413.

The effects of positive visualization with and without sound on sleep actigraphy measures in a small sample of tinnitus patients

BackgroundTinnitus is a common oto-neurological disorder associated with sleep disturbances. This research explored visualization and visualization reinforced with nature sounds as an acute therapy for sleep disturbance in people with tinnitus.MethodsTwelve adults with bothersome tinnitus and sleep disturbances participated in a randomized cross-over design in which measurements of tinnitus perception (rating scales), anxiety/depression (Hospital Anxiety and Depression Scale), attention (Attention and Performance Self-Assessment), and sleep quality (sleep diary and actigraphy—automated estimates of total sleep time, sleep onset latency, sleep fragmentation, and wake after sleep onset) were undertaken at baseline and repeated with therapy. The visualization task was 30 min in duration and was performed every night before sleep. The visualization with sound reinforcement task added nature sounds played overnight.ResultsSleep onset latency analyzed using the sleep diary normalized actigraphy was significantly reduced with both interventions (visualization reduced by 60 min and visualization with sound reinforcement reduced by 70 min). None of the other sleep quality markers demonstrated a statistically significant change. Self-reported attention ability significantly improved with visualization. Of the participants, 90% were able to consistently complete actigraphy measures over the 3 weeks of the study, and 75% kept sleep diaries.ConclusionSleep onset time was reduced with visualization.

Non-pharmacological interventions for sleep disruptions and fatigue after traumatic brain injury: a scoping review.

The aim of this study was to conduct a scoping review to determine the nature, variety, and volume of empirical evidence on nonpharmacological interventions for sleep disturbances with potential implications for fatigue in adults sustaining a traumatic brain injury (TBI). A systematic literature search was conducted across four databases to identify primary studies testing a single non-pharmacological intervention or a combination of non-pharmacological interventions for sleep disturbances and fatigue in community-dwelling adults with TBI. Sixteen studies were reviewed addressing six non-pharmacological interventions for sleep disruptions and fatigue after TBI including light therapy, cognitive-behavioral therapy, warm footbath application, shiatsu, and sleep hygiene protocol. Non-pharmacological interventions involving light or cognitive-behavioral therapy were reported in 75% of the studies. Actigraphy-based estimation of total sleep time and subjective level of fatigue were frequent outcomes. While this scoping review has utility in describing existing non-pharmacological approaches to manage sleep and fatigue after TBI, the findings suggest that interventions are often developed without considering TBI individuals' source of motivation and the need for support in self-administration. Future studies may achieve greater sustainability by considering the evolving needs of TBI patients and their families and the drivers and barriers that might influence non-pharmacological intervention use at home.

Sleep and dizziness in middle-aged and elderly persons: A cross-sectional population-based study

ObjectivesDizziness is a physical and mental burden that affects 33% of the population at age 70. Research in clinical samples suggests that poor sleep may contribute to dizziness complaints, but this has yet to be determined in the general population. We assessed the association of self-reported and actigraphy-estimated sleep with dizziness in a population-based sample. MethodsData of 4702 participants from the population-based Rotterdam Study, included between 2011 and 2014, were analyzed (mean age: 65.8 years, 55.7% women). Sleep quality was measured with the Pittsburgh Sleep Quality Index. Experience of dizziness and associated characteristics were self-reported, characteristics were categorized as vestibular or non-vestibular. A subsample of 1440 participants (mean age: 64.9 years, 51.3% women) wore an actigraph for a median of 7 days to allow objective estimation of total sleep time, sleep efficiency, sleep onset latency, and wake after sleep onset. We used logistic regression to assess the associations between different sleep parameters and dizziness. ResultsPoor self-reported sleep quality was associated with more dizziness (OR = 1.065, 95% CI: 1.043 to 1.087). This effect size was somewhat larger for dizziness with non-vestibular characteristics (OR = 1.108, 1.069 to 1.149) than for dizziness with vestibular characteristics (OR = 1.062, 0.988 to 1.142). Actigraphy-estimated sleep parameters were not associated with dizziness. ConclusionsA poor self-reported sleep quality is associated with more dizziness in a population-based sample of middle-aged and elderly persons. We encourage clinicians to assess sleep quality in patients presenting with dizziness, and vice versa.

An explainable deep-learning model to stage sleep states in children and propose novel EEG-related patterns in sleep apnea

Automatic deep-learning models used for sleep scoring in children with obstructive sleep apnea (OSA) are perceived as black boxes, limiting their implementation in clinical settings. Accordingly, we aimed to develop an accurate and interpretable deep-learning model for sleep staging in children using single-channel electroencephalogram (EEG) recordings. We used EEG signals from the Childhood Adenotonsillectomy Trial (CHAT) dataset (n = 1637) and a clinical sleep database (n = 980). Three distinct deep-learning architectures were explored to automatically classify sleep stages from a single-channel EEG data. Gradient-weighted Class Activation Mapping (Grad-CAM), an explainable artificial intelligence (XAI) algorithm, was then applied to provide an interpretation of the singular EEG patterns contributing to each predicted sleep stage. Among the tested architectures, a standard convolutional neural network (CNN) demonstrated the highest performance for automated sleep stage detection in the CHAT test set (accuracy = 86.9% and five-class kappa = 0.827). Furthermore, the CNN-based estimation of total sleep time exhibited strong agreement in the clinical dataset (intra-class correlation coefficient = 0.772). Our XAI approach using Grad-CAM effectively highlighted the EEG features associated with each sleep stage, emphasizing their influence on the CNN's decision-making process in both datasets. Grad-CAM heatmaps also allowed to identify and analyze epochs within a recording with a highly likelihood to be misclassified, revealing mixed features from different sleep stages within these epochs. Finally, Grad-CAM heatmaps unveiled novel features contributing to sleep scoring using a single EEG channel. Consequently, integrating an explainable CNN-based deep-learning model in the clinical environment could enable automatic sleep staging in pediatric sleep apnea tests.

Identification of Sleep Patterns via Clustering of Hypnodensities.

Sleep patterns vary widely between individuals. We explore methods for identifying populations exhibiting similar sleep patterns in an automated fashion using polysomnography data. Our novel approach applies unsupervised machine learning algorithms to hypnodensities graphs generated by a pre-trained neural network. In a population of 100 subjects we identify two stable clusters whose characteristics we visualize graphically and through estimates of total sleep time. We also find that the hypnodensity representation of the sleep stages produces more robust clustering results than the same methods applied to traditional hypnograms.

The Difference in Spectral Power Density in Sleep Electroencephalography According to the Estimation of Total Sleep Time

Objective: The discrepancy between self-reported subjective sleep time and objective sleep time varies from person to person, and the mechanism that can explain this discrepancy is unclear. This study aimed to investigate the difference in the spectral power of sleep electroencephalography (EEG) according to the estimation of total sleep time (TST).Methods: Of the 4,080 participants in the Sleep Heart Health Study, 2,363 participants with complete data from polysomnography, morning sleep survey, and spectral power of sleep EEG were included in the analysis. The participants were classified as normo-estimators (estimation of TST <±60 min), under-estimators (underestimation of TST ≤60 min), or over-estimators (overestimation of TST ≥60 min). A fast Fourier transformation was used to calculate the EEG power spectrum for total sleep duration within contiguous 30-s epochs of sleep. The sleep EEG spectral power was compared among the groups after adjusting for potential confounding factors, such as age, sex, proportion of participants with insomnia, apnea-hypopnea index, and TST.Results: Of the 2,363 participants, 1,507 (63.8%), 412 (17.4%), and 444 (18.8%) were assigned to the normo-, under-, and over-estimator groups, respectively. The power spectra during total sleep differed significantly among the groups in the delta (p=0.008) and theta bands (p=0.017) after controlling for potential confounders.Conclusion: Higher delta and theta powers were found in the under-estimators than in the over-estimators. This study suggests that differences exist in the microstructures of polysomnography-derived sleep EEG between these two groups. This study suggests that differences exist in the microstructures of polysomnography-derived sleep EEG based on differences in sleep time estimation.

Comparison of sleep parameters from wrist-worn ActiGraph and Actiwatch devices.

Sleep and physical activity, two important health behaviors, are often studied independently using different accelerometer types and body locations. Understanding whether accelerometers designed for monitoring each behavior can provide similar sleep parameter estimates may help determine whether one device can be used to measure both behaviors. Three hundred and thirty one adults (70.7 ± 13.7 years) from the Baltimore Longitudinal Study of Aging wore the ActiGraph GT9X Link and the Actiwatch 2 simultaneously on the non-dominant wrist for 7.0 ± 1.6 nights. Total sleep time (TST), wake after sleep onset (WASO), sleep efficiency, number of wake bouts, mean wake bout length, and sleep fragmentation index (SFI) were extracted from ActiGraph using the Cole-Kripke algorithm and from Actiwatch using the software default algorithm. These parameters were compared using paired t-tests, Bland-Altman plots, and Deming regression models. Stratified analyses were performed by age, sex, and body mass index (BMI). Compared to the Actiwatch, the ActiGraph estimated comparable TST and sleep efficiency, but fewer wake bouts, longer WASO, longer wake bout length, and higher SFI (all p < .001). Both devices estimated similar 1-min and 1% differences between participants for TST and SFI (β = 0.99, 95% CI: 0.95, 1.03, and 0.91, 1.13, respectively), but not for other parameters. These differences varied by age, sex, and/or BMI. The ActiGraph and the Actiwatch provide comparable absolute and relative estimates of TST, but not other parameters. The discrepancies could result from device differences in movement collection and/or sleep scoring algorithms. Further comparison and calibration is required before these devices can be used interchangeably.

Pre-Sleep Cognitive Arousal Is Unrelated to Sleep Misperception in Healthy Sleepers When Unexpected Sounds Are Played during Non-Rapid Eye Movement Sleep: A Polysomnography Study.

Background: It is well-established that environmental noise can disrupt sleep, and cause a mismatch between subjective and objective sleep, which is known as “sleep misperception”. Naturalistic studies indicate that pre-sleep cognitive arousal and sleep misperception are associated in the context of noise. However, it is not known if this is the case when ecologically valid noises are specifically played during non-rapid eye movement (NREM) sleep, which is susceptible to noise-related disruption. The present study evaluated if pre-sleep cognitive arousal was associated with sleep misperception in healthy normal sleepers, when unexpected ecologically valid common nocturnal noises were played during NREM sleep. Methods: Eighteen healthy sleepers (Mage = 23.37 years, SDage = 3.21 years) participated. Sleep was measured objectively on three consecutive nights using polysomnography, in a sleep laboratory environment, and subjectively, through participant estimates of total sleep time (TST). Night 1 was a baseline night where no noises were played. On Night 2, noises, which were chosen to be representative of habitual nocturnal noises heard in home environments, were played to participants via in-ear headphones after 5 min of objective sleep. Results: Unexpectedly, habitual pre-sleep cognitive arousal was not associated with subjective–objective TST discrepancy on Night 2. Conclusions: These results suggest that in healthy sleepers, when ecologically valid noises are played unexpectedly during NREM sleep in an unfamiliar sleep laboratory environment the subjective experience of sleep is not associated with pre-sleep cognitive arousal, or negatively impacted by noise exposure.

Does Sleep Quality Influence Recovery Outcomes After Postconcussive Injury in Children and Adolescents?

To determine whether objective parameters of sleep quality differ throughout recovery between children and adolescents who experienced an early return to school (RTS) and those who had a delayed RTS or did not return at all during the study period. Sleep parameters reflective of sleep quality were evaluated in participants' natural sleeping habitat throughout 9 weeks postinjury. Ninety-four children and adolescents (aged 5-18 years) with diagnosed concussion. Prospective cohort. Participants followed RTS protocols. Actigraphy-derived estimates of total sleep time (TST), sleep efficiency (SE), wake after sleep onset (WASO), average arousal length (AAL), and number of arousals (NOAs) per hour were assessed. The length of time from injury until RTS was determined for each participant. Participants were categorized into an early RTS or delayed RTS group based on their time to RTS. Both TST and SE were significantly greater in the early RTS group. WASO duration, AAL, and NOAs were significantly greater in the delayed RTS group. Differences between RTS groups were most apparent during weeks 1 to 5 postinjury. Participants who returned to school earlier had significantly better objective sleep quality than participants who experienced a delayed RTS. This study provides evidence in support of a relationship between sleep quality and time to RTS in children and adolescents with concussion. Considering early monitoring of sleep, education regarding sleep hygiene, and access to age-appropriate sleep interventions may be helpful in pediatric concussion recovery.

0139 Work Duration Affects How Prior-Night Sleep Predicts Next-Day Energy Expenditure in Emergency Response System Telecommunicators

Abstract Introduction ERS telecommunicators are the first of the first responders challenged with solving complex, time-sensitive problems while managing workplace presence. Very little is known about sleep, work, and lifestyle factors among workers in this industry. One study demonstrated that 85% of ERS telecommunicators are overweight, suggesting that job-related factors may place these workers at risk for sedentary lifestyles. To test this hypothesis, we examined whether 14 day total work duration moderated the daily relationship between prior-night total sleep time and next day energy expenditure. Methods Over the course of 14 days (M = 6.9 days on-shift; SD = 1.9 days), 47 ERS telecommunicators were instructed to (a) wear actigraphs on their waist to gather estimates of average energy expenditure (EE, kcal/hour), (b) wear actigraphs on their wrist to gather estimates of total sleep time (min), and (c) complete daily shift logs to gather information about work duration (hours). Mixed linear modeling was employed to examine whether prior night within-subject total sleep time (TST) predicted next day energy expenditure, as moderated by between-subject work hours (n = 525 cases). Results A significant cross-level Work Duration x TST interaction (Estimate = .007, SE = .002, p &amp;lt; .001, 95% CI [.003, .011]) indicated that less prior-night TST was associated with less next-day EE among telecommunicators who worked more hours over the last 14 days. Conversely, telecommunicators who worked fewer hours expended more energy per hour the next day when they slept less than usual. Simple effects indicated that for each extra 102 minutes sleep (+1 SD), telecommunicators expended 5 kcal/hr (90 kcal over 18 hours awake). These results remained stable when controlling for between-subject differences in sleep and within-subject changes in work duration, night-shift work, and other relevant covariates. Conclusion The effect of total sleep time on next-day EE is unique to each telecommunicator’s typical sleep levels and the total hours worked over the course of two weeks. These two risk factors operate on EE as a function of one another. Findings provide support for the implementation of policy-level intervention to minimize chronic overwork and individual-level intervention to support sleep prioritization. Support (If Any) UA Canyon Ranch Center for Health Promotion and Treatment

Association Between Subjective-Objective Discrepancy of Sleeping Time and Health-Related Quality of Life: A Community-Based Polysomnographic Study.

This study aimed to investigate the existence of a difference in quality of life (QOL) between individuals with and without significant subjective-objective discrepancy (SOD) in total sleep time (TST). From the Sleep Heart Health Study 2, 2540 individuals who had completed polysomnography, a morning sleep survey, and the 36-item Short-Form Health Survey (SF-36) were included in the analyses. The participants were classified as normoestimators (estimation of TST <±60 minutes), underestimators (underestimation of TST ≥60 minutes), or overestimators (overestimation of TST ≥60 minutes). The standardized SF-36 QOL scores were compared among the three groups. An adjusted partial correlation analysis was conducted between SOD and QOL. Of the 2540 participants, 1617 (63.7%), 433 (17.0%), and 490 (19.3%) were assigned to the normoestimator, underestimator, and overestimator groups, respectively. The bodily pain and social functioning components of the SF-36 score were significantly lower in the underestimators than in the normoestimators, whereas the physical functioning component was significantly lower in the overestimators than in the normoestimators. The absolute value of SOD in the TST showed a significant negative correlation with the physical and mental components of the SF-36. QOL was significantly better in the normoestimator than in the other groups and linearly correlated with the absolute value of SOD. This study suggests that a high prevalence of positive and negative sleep misperception in a community population can be a potential factor associated with poor QOL and potential comorbidities.

Minimum number of nights for reliable estimation of habitual sleep using a consumer sleep tracker

To determine the minimum number of nights required to reliably estimate weekly and monthly mean sleep duration and sleep variability measures from a consumer sleep technology (CST) device (Fitbit). Data comprised 107 144 nights from 1041 working adults aged 21-40 years. Intraclass correlation (ICC) analyses were conducted on both weekly and monthly time windows to determine the number of nights required to achieve ICC values of 0.60 and 0.80, corresponding to "good" and "very good" reliability thresholds. These minimum numbers were then validated on data collected 1-month and 1-year later. Minimally, 3 and 5 nights were required to obtain "good" and "very good" mean weekly total sleep time (TST) estimates, while 5 and 10 nights were required for monthly TST estimates. For weekday-only estimates, 2 and 3 nights were sufficient for weekly time windows while 3 and 7 nights sufficed for monthly time windows. Weekend-only estimates of monthly TST required 3 and 5 nights. TST variability required 5 and 6 nights for weekly time windows, and 11 and 18 nights for monthly time windows. Weekday-only weekly variability required 4 nights for both "good" and "very good" estimates while monthly variability required 9 and 14 nights. Weekend-only estimates of monthly variability required 5 and 7 nights. Error estimates made using data collected 1-month and 1-year later with these parameters were comparable to those associated with the original dataset. Studies should consider the metric, measurement window of interest, and desired reliability threshold to decide on the minimum number of nights required to assess habitual sleep using CST devices.

Sleep Diagnostics for Home Monitoring of Sleep Apnea Patients.

Objectives: Sleep time information is essential for monitoring of obstructive sleep apnea (OSA), as the severity assessment depends on the number of breathing disturbances per hour of sleep. However, clinical procedures for sleep monitoring rely on numerous uncomfortable sensors, which could affect sleeping patterns. Therefore, an automated method to identify sleep intervals from unobtrusive data is required. However, most unobtrusive sensors suffer from data loss and sensitivity to movement artifacts. Thus, current sleep detection methods are inadequate, as these require long intervals of good quality. Moreover, sleep monitoring of OSA patients is often less reliable due to heart rate disturbances, movement and sleep fragmentation. The primary aim was to develop a sleep-wake classifier for sleep time estimation of suspected OSA patients, based on single short-term segments of their cardiac and respiratory signals. The secondary aim was to define metrics to detect OSA patients directly from their predicted sleep-wake pattern and prioritize them for clinical diagnosis.Methods: This study used a dataset of 183 suspected OSA patients, of which 36 test subjects. First, a convolutional neural network was designed for sleep-wake classification based on healthier patients (AHI < 10). It employed single 30 s epochs of electrocardiograms and respiratory inductance plethysmograms. Sleep information and Total Sleep Time (TST) was derived for all patients using the short-term segments. Next, OSA patients were detected based on the average confidence of sleep predictions and the percentage of sleep-wake transitions in the predicted sleep architecture.Results: Sleep-wake classification on healthy, mild and moderate patients resulted in moderate κ scores of 0.51, 0.49, and 0.48, respectively. However, TST estimates decreased in accuracy with increasing AHI. Nevertheless, severe patients were detected with a sensitivity of 78% and specificity of 89%, and prioritized for clinical diagnosis. As such, their inaccurate TST estimate becomes irrelevant. Excluding detected OSA patients resulted in an overall estimated TST with a mean bias error of 21.9 (± 55.7) min and Pearson correlation of 0.74 to the reference.Conclusion: The presented framework offered a realistic tool for unobtrusive sleep monitoring of suspected OSA patients. Moreover, it enabled fast prioritization of severe patients for clinical diagnosis.

262 Comparing Two Methods for Scoring Wrist Actigraphy to Polysomnography for Estimating Total Sleep Time in Healthy Men

Abstract Introduction This analysis assessed whether manually setting rest (i.e., time in bed) intervals prior to using a proprietary software package (Actiware, version 6.09) to analyze wrist actigraphy data improved estimates of total sleep time (TST) compared to polysomnography (PSG). Methods The Phillips Actiwatch 2 and PSG (reference method) were used to calculate TST on two separate nights in twelve men (age=28.3 ± 5.7). Participants had an 8-hour sleep opportunity on night one and a 5-hour sleep opportunity and on night two. Estimates of TST from actigraphy data were calculated using two scoring methods. For scoring method 1, we allowed the software to automatically choose rest intervals and then applied a proprietary algorithm to calculate TST. For scoring method 2, we manually entered rest intervals using a published decision tree that incorporates activity, light, event marker, and sleep diary data. After the rest intervals were set in method 2, the proprietary algorithm was applied to calculate TST. Mean bias and limits of agreement (LOA) from Bland-Altman plots compared TST derived from both actigraphy scoring methods to PSG estimates. Results On night 1 (n=8) TST measured by PSG was 398.4 ± 40.6 minutes, compared to 395.5 ± 70.9 minutes using actigraphy scoring method 1 and 396 ± 44.5 minutes using scoring method 2. Mean bias was similar when comparing both scoring methods to PSG, but the LOA were wider in method 1 compared to method 2 (method 1 vs. PSG: -2.9 [-110.4, 104.7]; method 2 vs. PSG: -2.4 [-66.5, 61.7]; minutes). On night 2 (n=12) TST determined by PSG was 283.3 ± 11.2 minutes, compared to 302.1 ± 84.4 minutes using actigraphy scoring method 1 and 273.1 ± 14.5 minutes using scoring method 2. Again, LOA for TST estimated by actigraphy scoring method 1 were wider compared to scoring method number 2 (method 1 vs. PSG: 18.8 [-136.9, 174.6]; method 2 vs. PSG: -10.2 [-35.1, 14.8]). Conclusion These data demonstrate that applying a decision tree to manually set time in bed intervals prior to running analyses in the software results in better agreement when estimating TST from wrist actigraphy compared to PSG. Support (if any) UL1RR025780, K23AR070275.

0445 A Novel Algorithm for the Estimation of Sleep States Based on Breathing and Movement

Abstract Introduction We tested the diagnostic accuracy of the novel Nox BodySleep™ algorithm (Nox Medical, Iceland) for the estimation of sleep states from polygraphy (PG) sleep recordings based on features extracted from actigraphy and respiratory inductance plethysmography (RIP) belts. The algorithm automatically classifies epochs into three states, Wake, REM sleep and NonREM sleep. Validation was performed against polysomnography (PSG) in a sleep laboratory collective including patients with sleep disordered breathing (SBAS) and sleep related movements disorders. Methods Patients received PSG according to clinical routine. The recording was evaluated by the novel algorithm and the results were evaluated by descriptive statistics methods (IBM SPSS Statistics 25.0). Results We found a good Spearman correlation (r=0.8) and a bias of 11 minutes for the estimation of Total Sleep Time. Sleep Efficiency was also valued with a good Spearman correlation (r=0.7) and a bias of 1.6%. Wake phases were estimated with a F1 score of 0.64 while REM and Non-REM phases were evaluated with a F1 score of 0.73 and 0.82, respectively. Additionally, an overall accuracy of 0.8 and a Cohens kappa of 0.7 were found. Patients with sleep related movement disorders showed a slighly weaker correlation as patients with SBAS. Conclusion The algorithm shows a good diagnostic accuracy for the estimation of sleep states and significant sleep parameters. After validation on a larger patient collective, it could be used in the ambulatory and telemedical field to allow investigations comparable to the accuracy of a PSG. Support No support.

A comparison of automated and manual sleep staging and respiratory event recognition in a portable sleep diagnostic device with in-lab sleep study.

The objectives were to develop and validate an algorithm for editing WatchPAT scoring and assess the accuracy in an unselected clinical population as well as age and sex substrata. Two hundred sixty-two participants were enrolled to undergo WatchPAT simultaneously with in-lab polysomnography (PSG) recordings for developing (n = 30), optimizing (n = 62), and validating (n = 170) an algorithm to review and edit respiratory events and sleep architecture of WatchPAT recordings, which was based on visual inspection of WatchPAT signals. Apnea-hypopnea index (AHI) and sleep indices were compared with PSG-derived and automated WatchPAT indices. Although estimation of total sleep time (TST) was comparable between automated and manual algorithm, estimation of rapid eye movement (REM) sleep time was markedly improved with manual editing from 0.48, 23.0 min (-43.9 to 89.8) to 0.64, 18.3 min (-32.6 to 69.1) (correlation with PSG, mean difference [reference range] from PSG, respectively). Manual scoring also improved correlation and agreement with PSG AHI from 0.65, 2.5 events/h (-24.0 to 28.9) to 0.81, -4.5 events/h (-22.5 to 13.6) as well as concordance for categorical agreement of sleep-disordered breathing severity and concordance for detecting severe REM-related sleep-disordered breathing. Interscorer reliabilities were excellent for TST and AHI, while good for REM sleep time. The automated algorithm performed better in younger than in older patients, while performed similarly between men and women with respect to concordance statistics. The manual algorithm markedly improved concordances more in older patients and women than in their counterparts. Our manual editing algorithm improves correlation and agreement with PSG-derived sleep and breathing indices. Sex and age influence the accuracy of automated analysis and the performance of manual editing on AHI concordance.

Personalized Sleep Parameters Estimation from Actigraphy: A Machine Learning Approach.

BackgroundThe current gold standard for measuring sleep is polysomnography (PSG), but it can be obtrusive and costly. Actigraphy is a relatively low-cost and unobtrusive alternative to PSG. Of particular interest in measuring sleep from actigraphy is prediction of sleep-wake states. Current literature on prediction of sleep-wake states from actigraphy consists of methods that use population data, which we call generalized models. However, accounting for variability of sleep patterns across individuals calls for personalized models of sleep-wake states prediction that could be potentially better suited to individual-level data and yield more accurate estimation of sleep.PurposeTo investigate the validity of developing personalized machine learning models, trained and tested on individual-level actigraphy data, for improved prediction of sleep-wake states and reliable estimation of nightly sleep parameters.Participants and methodsWe used a dataset including 54 participants and systematically trained and tested 5 different personalized machine learning models as well as their generalized counterparts. We evaluated model performance compared to concurrent PSG through extensive machine learning experiments and statistical analyses.ResultsOur experiments show the superiority of personalized models over their generalized counterparts in estimating PSG-derived sleep parameters. Personalized models of regularized logistic regression, random forest, adaptive boosting, and extreme gradient boosting achieve estimates of total sleep time, wake after sleep onset, sleep efficiency, and number of awakenings that are closer to those obtained by PSG, in absolute difference, than the same estimates from their generalized counterparts. We further show that the difference between estimates of sleep parameters obtained by personalized models and those of PSG is statistically non-significant.ConclusionPersonalized machine learning models of sleep-wake states outperform their generalized counterparts in terms of estimating sleep parameters and are indistinguishable from PSG labeled sleep-wake states. Personalized machine learning models can be used in actigraphy studies of sleep health and potentially screening for some sleep disorders.

Optimizing Actigraphic Estimation of Sleep Duration in Suspected Idiopathic Hypersomnia.

To determine the optimal Actiwatch 2 setting configuration for the estimation of total sleep time (TST) in persons with suspected idiopathic hypersomnia. Thirty-three patients with a diagnosis of idiopathic hypersomnia (28 female; mean age = 33.7 ± 10.5) underwent ad libitum polysomnography with concurrent use of the Actiwatch 2. Actiwatch 2 sleep-wake activity threshold (SWAT; Low, Medium, and High) and sleep immobility onset and offset (SIOO; 5, 10, 15, 20, 25, and 30 epoch) duration were modified during data processing. The resultant 18 unique setting combinations were subsequently evaluated using Bland-Altman and epoch comparison analyses to determine optimal settings relative to polysomnography. Low SWAT + 25 Epoch SIOO displayed the least divergence from polysomnography (mean difference 3.4 minutes). Higher SWAT and lower SIOO increased sensitivity and accuracy, but at the expense of reducing specificity and the ability to accurately estimate TST. These results demonstrate that actigraphic settings should be carefully considered when estimating sleep duration. The Low + 25 Epoch configuration is indicated as most optimal for estimating TST in persons with suspected idiopathic hypersomnia. A commentary on this article appears in this issue on page 539.

Automatic identification of sleep and wakefulness using single-channel EEG and respiratory polygraphy signals for the diagnosis of obstructive sleep apnea.

Polysomnography (PSG) is necessary for the accurate estimation of total sleep time (TST) and the calculation of the apnea-hypopnea index (AHI). In type III home sleep apnea testing (HSAT), TST is overestimated because of the lack of electrophysiological sleep recordings. The aim of this study was to evaluate the accuracy and reliability of a novel automated sleep/wake scoring algorithm combining a single electroencephalogram (EEG) channel with actimetry and HSAT signals. The study included 160 patients investigated by PSG for suspected obstructive sleep apnea (OSA). Each PSG was recorded and scored manually using American Academy of Sleep Medicine (AASM) rules. The automatic sleep/wake-scoring algorithm was based on a single-channel EEG (FP2-A1) and the variability analysis of HSAT signals (airflow, snoring, actimetry, light and respiratory inductive plethysmography). Optimal detection thresholds were derived for each signal using a training set. Automatic and manual scorings were then compared epoch by epoch considering two states (sleep and wake). Cohen's kappa coefficient between the manual scoring and the proposed automatic algorithm was substantial, 0.74±0.18, in separating wakefulness and sleep. The sensitivity, specificity and the positive and negative predictive values for the detection of wakefulness were 76.51%±21.67%, 95.48%±5.27%, 81.84%±15.42% and 93.85%±6.23% respectively. Compared with HSAT signals alone, AHI increased by 22.12% and 27 patients changed categories of OSA severity with the automatic sleep/wake-scoring algorithm. Automatic sleep/wake detection using a single-channel EEG combined with HSAT signals was a reliable method for TST estimation and improved AHI calculation compared with HSAT.