The National Sleep Research Resource: towards a sleep data commons

Nocturnal oxygen resaturation parameters are associated with cardiorespiratory comorbidities

BACKGROUND

There are strong associations between oxygen desaturations and cardiovascular outcomes. Additionally, oxygen resaturation rates are linked to excessive daytime sleepiness independent of oxygen desaturation severity. No studies have yet looked at the independent effects of comorbidities or medications on resaturation parameters.

METHODS

The Sleep Heart Health Study data was utilised to derive oxygen saturation parameters from 5804 participants. Participants with a history of comorbidities or medication usage were compared against healthy participants with no comorbidity/medication history.

RESULTS

4293 participants (50.4% female, median age 64 years) were included in the analysis. Females recorded significantly faster resaturation rates (mean 0.61%/s) than males (mean 0.57%/s, p < 0.001), regardless of comorbidities. After adjusting for demographics, sleep parameters, and desaturation parameters, resaturation rate was reduced with hypertension (-0.09 (95% CI -0.16, -0.03)), myocardial infarction (-0.13 (95% CI -0.21, -0.04)) and heart failure (-0.19 (95% CI -0.33, -0.05)), or when using anti-hypertensives (-0.10 (95% CI -0.17, -0.03)), mental health medications (-0.18 (95% CI -0.27, -0.08)) or anticoagulants (-0.41 (95% CI -0.56, -0.26)). Desaturation to Resaturation ratio for duration was decreased with mental health (-0.21 (95% CI -0.34, -0.08)) or diabetic medications (-0.24 (95% CI -0.41, -0.07)), and desaturation to resaturation ratio for area decreased with heart failure (-0.25 (95% CI -0.42, -0.08)).

CONCLUSIONS

Comorbidities and medications significantly affect nocturnal resaturation parameters, independent of desaturation parameters. However, the causal relationship remains unclear. Further research can enhance our knowledge and develop more precise and safer interventions for individuals affected by certain comorbidities.

Expert-level sleep staging using an electrocardiography-only feed-forward neural network

Reliable classification of sleep stages is crucial in sleep medicine and neuroscience research for providing valuable insights, diagnoses, and understanding of brain states. The current gold standard method for sleep stage classification is polysomnography (PSG). Unfortunately, PSG is an expensive and cumbersome process involving numerous electrodes, often conducted in an unfamiliar clinic and annotated by a professional. Although commercial devices like smartwatches track sleep, their performance is well below PSG. To address these disadvantages, we present a feed-forward neural network that achieves gold-standard levels of agreement using only a single lead of electrocardiography (ECG) data. Specifically, the median five-stage Cohen's kappa is 0.725 on a large, diverse dataset of 5 to 90-year-old subjects. Comparisons with a comprehensive meta-analysis of between-human inter-rater agreement confirm the non-inferior performance of our model. Finally, we developed a novel loss function to align the training objective with Cohen's kappa. Our method offers an inexpensive, automated, and convenient alternative for sleep stage classification-further enhanced by a real-time scoring option. Cardiosomnography, or a sleep study conducted with ECG only, could take expert-level sleep studies outside the confines of clinics and laboratories and into realistic settings. This advancement democratizes access to high-quality sleep studies, considerably enhancing the field of sleep medicine and neuroscience. It makes less-expensive, higher-quality studies accessible to a broader community, enabling improved sleep research and more personalized, accessible sleep-related healthcare interventions.

A Review on Automated Sleep Study

In recent years, research on automated sleep analysis has witnessed significant growth, reflecting advancements in understanding sleep patterns and their impact on overall health. This review synthesizes findings from an exhaustive analysis of 87 papers, systematically retrieved from prominent databases such as Google Scholar, PubMed, IEEE Xplore, and ScienceDirect. The selection criteria prioritized studies focusing on methods employed, signal modalities utilized, and machine learning algorithms applied in automated sleep analysis. The overarching goal was to critically evaluate the strengths and weaknesses of the proposed methods, shedding light on the current landscape and future directions in sleep research. An in-depth exploration of the reviewed literature revealed a diverse range of methodologies and machine learning approaches employed in automated sleep studies. Notably, K-Nearest Neighbors (KNN), Ensemble Learning Methods, and Support Vector Machine (SVM) emerged as versatile and potent classifiers, exhibiting high accuracies in various applications. However, challenges such as performance variability and computational demands were observed, necessitating judicious classifier selection based on dataset intricacies. In addition, the integration of traditional feature extraction methods with deep structures and the combination of different deep neural networks were identified as promising strategies to enhance diagnostic accuracy in sleep-related studies. The reviewed literature emphasized the need for adaptive classifiers, cross-modality integration, and collaborative efforts to drive the field toward more accurate, robust, and accessible sleep-related diagnostic solutions. This comprehensive review serves as a solid foundation for researchers and practitioners, providing an organized synthesis of the current state of knowledge in automated sleep analysis. By highlighting the strengths and challenges of various methodologies, this review aims to guide future research toward more effective and nuanced approaches to sleep diagnostics.

Hypoxic indices for obstructive sleep apnoea severity and cardiovascular disease risk prediction: A comparison and application in a community population

BACKGROUND AND OBJECTIVE

The apnoea-hypopnoea index (AHI) and oxygen desaturation index (ODI) encounter challenges in capturing the intricate relationship between obstructive sleep apnoea (OSA) and cardiovascular disease (CVD) risks. Although novel hypoxic indices have been proposed to tackle these limitations, there remains a gap in comprehensive validation and comparisons across a unified dataset.

METHODS

Samples were derived from the Sleep Heart Health Study (SHHS), involving 4485 participants aged over 40 years after data quality screening. The study compared several key indices, including AHI, ODI, the reconstructed hypoxic burden (rHB), the percentage of sleep time with the duration of respiratory events causing desaturation (pRED_3p) and the sleep breathing impairment index (SBII), in relation to CVD mortality and morbidity risks. Adjusted Cox proportional models were employed to calculate hazard ratios (HRs) for each index, and comparisons were performed.

RESULTS

SBII and pRED_3p exhibited significant correlations with both CVD mortality and morbidity, with SBII showing the highest adjusted HR (95% confidence interval) for mortality (2.04 [1.25, 3.34]) and pRED_3p for morbidity (1.43 [1.09-1.88]). In contrast, rHB was only significant in predicting CVD mortality (1.63 [1.05-2.53]), while AHI and ODI did not show significant correlations with CVD outcomes. The adjusted models based on SBII and pRED_3p exhibited optimal performance in the CVD mortality and morbidity datasets, respectively.

CONCLUSION

This study identified the optimal indices for OSA-related CVD risks prediction, SBII for mortality and pRED_3p for morbidity. The open-source online platform provides the computation of the indices.

Deciphering the Link: Correlating REM Sleep Patterns with Depressive Symptoms via Consumer Wearable Technology

This study investigates the correlation between REM sleep patterns, as measured by the Apple Watch, and depressive symptoms in an undiagnosed population. Employing the Apple Watch for data collection, REM sleep duration and frequency were monitored over a specified period. Concurrently, participants' depressive symptoms were evaluated using standardized questionnaires. The analysis, primarily using Spearman's correlation, revealed noteworthy findings. A significant correlation was observed between an increased REM sleep proportion and higher depressive symptom scores, with a correlation coefficient of 0.702, suggesting a robust relationship. These results highlight the potential of using wearable technology, such as the Apple Watch, in early detection and intervention for depressive symptoms, suggesting that alterations in REM sleep could serve as preliminary indicators of depressive tendencies. This approach offers a non-invasive and accessible means to monitor and potentially preempt the progression of depressive disorders. This study's implications extend to the broader context of mental health, emphasizing the importance of sleep assessment in routine health evaluations, particularly for individuals exhibiting early signs of depressive symptoms.

Association Between Sleep Quality and Deep Learning-Based Sleep Onset Latency Distribution Using an Electroencephalogram

To evaluate sleep quality, it is necessary to monitor overnight sleep duration. However, sleep monitoring typically requires more than 7 hours, which can be inefficient in termxs of data size and analysis. Therefore, we proposed to develop a deep learning-based model using a 30 sec sleep electroencephalogram (EEG) early in the sleep cycle to predict sleep onset latency (SOL) distribution and explore associations with sleep quality (SQ). We propose a deep learning model composed of a structure that decomposes and restores the signal in epoch units and a structure that predicts the SOL distribution. We used the Sleep Heart Health Study public dataset, which includes a large number of study subjects, to estimate and evaluate the proposed model. The proposed model estimated the SOL distribution and divided it into four clusters. The advantage of the proposed model is that it shows the process of falling asleep for individual participants as a probability graph over time. Furthermore, we compared the baseline of good SQ and SOL and showed that less than 10 minutes SOL correlated better with good SQ. Moreover, it was the most suitable sleep feature that could be predicted using early EEG, compared with the total sleep time, sleep efficiency, and actual sleep time. Our study showed the feasibility of estimating SOL distribution using deep learning with an early EEG and showed that SOL distribution within 10 minutes was associated with good SQ.

Development and validation of a novel sleep health score in the sleep heart health study

BACKGROUND

There is a lack of consensus in evaluating multidimensional sleep health, especially concerning its implication for mortality. A validated multidimensional sleep health score is the foundation of effective interventions.

METHODS

We obtained data from 5706 participants in the Sleep Heart Health Study. First, random forest-recursive feature elimination algorithm was used to select potential predictive variables. Second, a sleep composite score was developed based on the regression coefficients from a Cox proportional hazards model evaluating the associations between selected sleep-related variables and mortality. Last, we validated the score by constructing Cox proportional hazards models to assess its association with mortality.

RESULTS

The mean age of participants was 63.2 years old, and 47.6% (2715/5706) were male. Six sleep variables, including average oxygen saturation (%), spindle density (C3), sleep efficiency (%), spindle density (C4), percentage of fast spindles (%) and percentage of rapid eye movement (%) were selected to construct this multidimensional sleep health score. The average sleep composite score in participants was 6.8 of 22 (lower is better). Participants with a one-point increase in sleep composite score had an 10% higher risk of death (hazard ratio = 1.10, 95% confidence interval: 1.08-1.12).

CONCLUSIONS

This study constructed and validated a novel multidimensional sleep health score to better predict death based on sleep, with significant associations between sleep composite score and all-cause mortality. Integrating questionnaire information and sleep microstructures, our sleep composite score is more appropriately applied for mortality risk stratification.

SleepBoost: a multi-level tree-based ensemble model for automatic sleep stage classification

Neurodegenerative diseases often exhibit a strong link with sleep disruption, highlighting the importance of effective sleep stage monitoring. In this light, automatic sleep stage classification (ASSC) plays a pivotal role, now more streamlined than ever due to the advancements in deep learning (DL). However, the opaque nature of DL models can be a barrier in their clinical adoption, due to trust concerns among medical practitioners. To bridge this gap, we introduce SleepBoost, a transparent multi-level tree-based ensemble model specifically designed for ASSC. Our approach includes a crafted feature engineering block (FEB) that extracts 41 time and frequency domain features, out of which 23 are selected based on their high mutual information score (> 0.23). Uniquely, SleepBoost integrates three fundamental linear models into a cohesive multi-level tree structure, further enhanced by a novel reward-based adaptive weight allocation mechanism. Tested on the Sleep-EDF-20 dataset, SleepBoost demonstrates superior performance with an accuracy of 86.3%, F1-score of 80.9%, and Cohen kappa score of 0.807, outperforming leading DL models in ASSC. An ablation study underscores the critical role of our selective feature extraction in enhancing model accuracy and interpretability, crucial for clinical settings. This innovative approach not only offers a more transparent alternative to traditional DL models but also extends potential implications for monitoring and understanding sleep patterns in the context of neurodegenerative disorders. The open-source availability of SleepBoost's implementation at https://github.com/akibzaman/SleepBoost can further facilitate its accessibility and potential for widespread clinical adoption.

Predicting the impact of CPAP on brain health: A study using the sleep EEG-derived brain age index

OBJECTIVE

This longitudinal study investigated potential positive impact of CPAP treatment on brain health in individuals with obstructive sleep Apnea (OSA). To allow this, we aimed to employ sleep electroencephalogram (EEG)-derived brain age index (BAI) to quantify CPAP's impact on brain health and identify individually varying CPAP effects on brain aging using machine learning approaches.

METHODS

We retrospectively analyzed CPAP-treated (n = 98) and untreated OSA patients (n = 88) with a minimum 12-month follow-up of polysomnography. BAI was calculated by subtracting chronological age from the predicted brain age. To investigate BAI changes before and after CPAP treatment, we compared annual ΔBAI between CPAP-treated and untreated OSA patients. To identify individually varying CPAP effectiveness and factors influencing CPAP effectiveness, machine learning approaches were employed to predict which patient displayed positive outcomes (negative annual ΔBAI) based on their baseline clinical features.

RESULTS

CPAP-treated group showed lower annual ΔBAI than untreated (-0.6 ± 2.7 vs. 0.3 ± 2.6 years, p < 0.05). This BAI reduction with CPAP was reproduced independently in the Apnea, Bariatric surgery, and CPAP study cohort. Patients with more severe OSA at baseline displayed more positive annual ΔBAI (=accelerated brain aging) when untreated and displayed more negative annual ΔBAI (=decelerated brain aging) when CPAP-treated. Machine learning models achieved high accuracy (up to 86%) in predicting CPAP outcomes.

INTERPRETATION

CPAP treatment can alleviate brain aging in OSA, especially in severe cases. Sleep EEG-derived BAI has potential to assess CPAP's impact on brain health. The study provides insights into CPAP's effects and underscores BAI-based predictive modeling's utility in OSA management.

Laying the Foundation: Modern Transformers for Gold-Standard Sleep Analysis and Beyond

Accurate sleep assessment is critical to the practice of sleep medicine and sleep research. The recent availability of large quantities of publicly available sleep data, alongside recent breakthroughs in AI like transformer architectures, present novel opportunities for data-driven discovery efforts. Transformers are flexible neural networks that not only excel at classification tasks, but also can enable data-driven discovery through un- or self-supervised learning, which requires no human annotations to the input data. While transformers have been extensively used in supervised learning scenarios for sleep stage classification, they have not been fully explored or optimized in forms designed from the ground up for use in un- or self-supervised learning tasks in sleep. A necessary first step will be to study these models on a canonical benchmark supervised learning task (5-class sleep stage classification). Hence, to lay the groundwork for future data-driven discovery efforts, we evaluated optimizations of a transformer-based architecture that has already demonstrated substantial success in self-supervised learning in another domain (audio speech recognition), and trained it to perform the canonical 5-class sleep stage classification task, to establish foundational baselines in the sleep domain. We found that small transformer models designed from the start for (later) self-supervised learning can match other state-of-the-art automated sleep scoring techniques, while also providing the basis for future data-driven discovery efforts using large sleep data sets.

Physics-Informed Transfer Learning to Enhance Sleep Staging

OBJECTIVE

At-home sleep staging using wearable medical sensors poses a viable alternative to in-hospital polysomnography due to its lower cost and lower disruption to the daily lives of patients, especially in the case of long-term monitoring. Machine learning with wearables however is difficult due to the paucity of data from wearable sensors, making automation a challenge. Transfer learning from hospital polysomnograms can boost performance, but is still hindered by differences between wearable and in-hospital EEG resulting in part from differing electrode placement. We improve transfer learning performance by using electrophysiological models of a human head to generate synthetic EEG resembling EEG from a wearable sensor.

METHODS

The data generation method utilizes Low-Resolution Electromagnetic Tomography Analysis (LORETA). Real EEG from standard in- hospital recordings is first mapped to point currents within the brain using LORETA, after which the point currents are used to estimate EEG that would have been recorded using a wearable sensor at any given point on the head.

RESULTS

Augmenting source datasets with synthetic data statistically significantly boosted accuracy on a wearable sleep staging task from 80.8% to 81.3% on average, depending on the transfer learning parameters and data sources.

CONCLUSION

Machine learning performance can be improved using data synthesized using physical models.

SIGNIFICANCE

Our approach represents a new form of transfer learning and demonstrates that incorporating domain knowledge of electrophysiological modeling can improve machine learning results for sleep staging tasks. We expect this approach to be particularly useful for EEG data which is hard to collect, or which is obtained using unusual electrode configurations.

[Formula: see text]-[Formula: see text]: A Nonparametric Model for Neural Power Spectra Decomposition

The power spectra estimated from the brain recordings are the mixed representation of aperiodic transient activity and periodic oscillations, i.e., aperiodic component (AC) and periodic component (PC). Quantitative neurophysiology requires precise decomposition preceding parameterizing each component. However, the shape, statistical distribution, scale, and mixing mechanism of AC and PCs are unclear, challenging the effectiveness of current popular parametric models such as FOOOF, IRASA, BOSC, etc. Here, ξ- π was proposed to decompose the neural spectra by embedding the nonparametric spectra estimation with penalized Whittle likelihood and the shape language modeling into the expectation maximization framework. ξ- π was validated on the synthesized spectra with loss statistics and on the sleep EEG and the large sample iEEG with evaluation metrics and neurophysiological evidence. Compared to FOOOF, both the simulation presenting shape irregularities and the batch simulation with multiple isolated peaks indicated that ξ- π improved the fit of AC and PCs with less loss and higher F1-score in recognizing the centering frequencies and the number of peaks; the sleep EEG revealed that ξ- π produced more distinguishable AC exponents and improved the sleep state classification accuracy; the iEEG showed that ξ- π approached the clinical findings in peak discovery. Overall, ξ- π offered good performance in the spectra decomposition, which allows flexible parameterization using descriptive statistics or kernel functions. ξ- π is a seminal tool for brain signal decoding in fields such as cognitive neuroscience, brain-computer interface, neurofeedback, and brain diseases.

Sleep architecture and the incidence of depressive symptoms in middle-aged and older adults: A community-based study

BACKGROUND

Rapid eye movement (REM) sleep and three stages of non-REM (NREM) sleep comprise the full sleep cycle. The changes in sleep have been linked to depression risk. This study aimed to explore the association between sleep architecture and depressive symptoms.

METHODS

A total of 3247 participants from the Sleep Heart Health Study (SHHS) were included in this cohort study. REM and NREM sleep were monitored by in-home polysomnography at SHHS visit 1. Depressive symptoms was reported as the first occurrence between SHHS visits 1 and 2 (mean follow-up of 5.3 years). Multivariable logistic regression was used to investigate the relationship between sleep stages and depressive symptoms.

RESULTS

In total, 225 cases of depressive symptoms (6.9 %) were observed between SHHS visits 1 and 2. A significant linear association between NREM Stage 1 and depressive symptoms was found after adjusting for potential covariates. Multivariable logistic regression analysis showed that percentage in NREM Stage 1 was associated with the incidence of depressive symptoms (odds ratio [OR], 1.06; 95 % confidence interval [CI], 1.02-1.10; P = 0.001), as were time in NREM Stage 1 and depressive symptoms (OR, 1.02; 95 % CI, 1.01-1.03; P = 0.001). However, no significant association with depressive symptoms was found for other sleep stage.

LIMITATIONS

The specific follow-up time for depressive symptoms diagnosis was missing.

CONCLUSIONS

Increased time or percentage in NREM Stage 1 was associated with a higher risk of developing depressive symptoms. The early change in sleep architecture were important for incidence of depressive symptoms and warrants constant concerns.

An Ensemble Classification Model for Depression Based on Wearable Device Sleep Data

Depression is one of the most common mental disorders, with sleep disturbances as typical symptoms. With the popularity of wearable devices increasing in recent years, more and more people wear portable devices to track sleep quality. Based on this, we believe that depression detection through wearable sleep data is more intelligent and economical. However, the majority of wearable devices face the problem of missing data during the data collection process. Otherwise, most existing studies of depression identification focus on the utilization of complex data, making it difficult to generalize and susceptible to noise interference. To address these issues, we propose a systematic ensemble classification model for depression (ECD). For the missing data problem of wearable devices, we design an improved GAIN method to further control the generation range of interpolated values, which can achieve a more reasonable treatment of missing values. Compared with the original GAIN approach, the improved method shows a 28.56% improvement when using MAE as the metric. For depression recognition, we use ensemble learning to construct a depression classification model which combines five classification models, including SVM, KNN, LR, CBR, and DT. Ensemble learning can improve the model's robustness and generalization. The voting mechanism is used in several places to improve noise immunity. The final classification model performed great on the dataset, with a precision of 92.55% and a recall of 91.89%. These results illustrate how efficient this method is in automatically detecting depression.

Self-Supervised Transformer Model Training for a Sleep-EEG Foundation Model

The American Academy of Sleep Medicine (AASM) recognizes five sleep/wake states (Wake, N1, N2, N3, REM), yet this classification schema provides only a high-level summary of sleep and likely overlooks important neurological or health information. New, data-driven approaches are needed to more deeply probe the information content of sleep signals. Here we present a self-supervised approach that learns the structure embedded in large quantities of neurophysiological sleep data. This masked transformer training procedure is inspired by high performing self-supervised methods developed for speech transcription. We show that self-supervised pre-training matches or outperforms supervised sleep stage classification, especially when labeled data or compute-power is limited. Perhaps more importantly, we also show that our pretrained model is flexible and can be fine-tuned to perform well on new tasks including distinguishing individuals and quantifying "brain age" (a potential health biomarker). This suggests that modern methods can automatically learn information that is potentially overlooked by the 5-class sleep staging schema, laying the groundwork for new schemas and further data-driven exploration of sleep.

Item response theory analysis of daytime sleepiness as a symptom of obstructive sleep apnea

Obstructive sleep apnea (OSA) is a sleep disorder which is linked to many health risks. The gold standard to evaluate OSA in clinical trials is the Apnea-Hypopnea Index (AHI). However, it is time-consuming, costly, and disregards aspects such as quality of life. Therefore, it is of interest to use patient-reported outcomes like the Epworth Sleepiness Scale (ESS), which measures daytime sleepiness, as surrogate end points. We investigate the link between AHI and ESS, via item response theory (IRT) modeling. Through the developed IRT model it was identified that AHI and ESS are not correlated to any high degree and probably not measuring the same sleepiness construct. No covariate relationships of clinical relevance were found. This suggests that ESS is a poor choice as an end point for clinical development if treatment is targeted at improving AHI, and especially so in a mild OSA patient group.

Association of Disrupted Delta Wave Activity During Sleep With Long-Term Cardiovascular Disease and Mortality

BACKGROUND

Delta wave activity is a prominent feature of deep sleep, which is significantly associated with sleep quality.

OBJECTIVES

The authors hypothesized that delta wave activity disruption during sleep could predict long-term cardiovascular disease (CVD) and CVD mortality risk.

METHODS

The authors used a comprehensive power spectral entropy-based method to assess delta wave activity during sleep based on overnight polysomnograms in 4,058 participants in the SHHS (Sleep Heart Health Study) and 2,193 participants in the MrOS (Osteoporotic Fractures in Men Study) Sleep study.

RESULTS

During 11.0 ± 2.8 years of follow-up in SHHS, 729 participants had incident CVD and 192 participants died due to CVD. During 15.5 ± 4.4 years of follow-up in MrOS, 547 participants had incident CVD, and 391 died due to CVD. In multivariable Cox regression models, lower delta wave entropy during sleep was associated with higher risk of coronary heart disease (SHHS: HR: 1.46; 95% CI: 1.02-2.06; P = 0.03; MrOS: HR: 1.79; 95% CI: 1.17-2.73; P < 0.01), CVD (SHHS: HR: 1.60; 95% CI: 1.21-2.11; P < 0.01; MrOS: HR: 1.43; 95% CI: 1.00-2.05; P = 0.05), and CVD mortality (SHHS: HR: 1.94; 95% CI: 1.18-3.18; P < 0.01; MrOS: HR: 1.66; 95% CI: 1.12-2.47; P = 0.01) after adjusting for covariates. The Shapley Additive Explanations method indicates that low delta wave entropy was more predictive of coronary heart disease, CVD, and CVD mortality risks than conventional sleep parameters.

CONCLUSIONS

The results suggest that delta wave activity disruption during sleep may be a useful metric to identify those at increased risk for CVD and CVD mortality.

The National Sleep Research Resource: making data findable, accessible, interoperable, reusable and promoting sleep science

This paper presents a comprehensive overview of the National Sleep Research Resource (NSRR), a National Heart Lung and Blood Institute-supported repository developed to share data from clinical studies focused on the evaluation of sleep disorders. The NSRR addresses challenges presented by the heterogeneity of sleep-related data, leveraging innovative strategies to optimize the quality and accessibility of available datasets. It provides authorized users with secure centralized access to a large quantity of sleep-related data including polysomnography, actigraphy, demographics, patient-reported outcomes, and other data. In developing the NSRR, we have implemented data processing protocols that ensure de-identification and compliance with FAIR (Findable, Accessible, Interoperable, Reusable) principles. Heterogeneity stemming from intrinsic variation in the collection, annotation, definition, and interpretation of data has proven to be one of the primary obstacles to efficient sharing of datasets. Approaches employed by the NSRR to address this heterogeneity include (1) development of standardized sleep terminologies utilizing a compositional coding scheme, (2) specification of comprehensive metadata, (3) harmonization of commonly used variables, and (3) computational tools developed to standardize signal processing. We have also leveraged external resources to engineer a domain-specific approach to data harmonization. We describe the scope of data within the NSRR, its role in promoting sleep and circadian research through data sharing, and harmonization of large datasets and analytical tools. Finally, we identify opportunities for approaches for the field of sleep medicine to further support data standardization and sharing.

ZleepAnlystNet: a novel deep learning model for automatic sleep stage scoring based on single-channel raw EEG data using separating training

Numerous models for sleep stage scoring utilizing single-channel raw EEG signal have typically employed CNN and BiLSTM architectures. While these models, incorporating temporal information for sequence classification, demonstrate superior overall performance, they often exhibit low per-class performance for N1-stage, necessitating an adjustment of loss function. However, the efficacy of such adjustment is constrained by the training process. In this study, a pioneering training approach called separating training is introduced, alongside a novel model, to enhance performance. The developed model comprises 15 CNN models with varying loss function weights for feature extraction and 1 BiLSTM for sequence classification. Due to its architecture, this model cannot be trained using an end-to-end approach, necessitating separate training for each component using the Sleep-EDF dataset. Achieving an overall accuracy of 87.02%, MF1 of 82.09%, Kappa of 0.8221, and per-class F1-socres (W 90.34%, N1 54.23%, N2 89.53%, N3 88.96%, and REM 87.40%), our model demonstrates promising performance. Comparison with sleep technicians reveals a Kappa of 0.7015, indicating alignment with reference sleep stags. Additionally, cross-dataset validation and adaptation through training with the SHHS dataset yield an overall accuracy of 84.40%, MF1 of 74.96% and Kappa of 0.7785 when tested with the Sleep-EDF-13 dataset. These findings underscore the generalization potential in model architecture design facilitated by our novel training approach.

A perspective on automated rapid eye movement sleep assessment

Rapid eye movement sleep is associated with distinct changes in various biomedical signals that can be easily captured during sleep, lending themselves to automated sleep staging using machine learning systems. Here, we provide a perspective on the critical characteristics of biomedical signals associated with rapid eye movement sleep and how they can be exploited for automated sleep assessment. We summarise key historical developments in automated sleep staging systems, having now achieved classification accuracy on par with human expert scorers and their role in the clinical setting. We also discuss rapid eye movement sleep assessment with consumer sleep trackers and its potential for unprecedented sleep assessment on a global scale. We conclude by providing a future outlook of computerised rapid eye movement sleep assessment and the role AI systems may play.

Sleep stage prediction using multimodal body network and circadian rhythm

Quality sleep plays a vital role in living beings as it contributes extensively to the healing process and the removal of waste products from the body. Poor sleep may lead to depression, memory deficits, heart, and metabolic problems, etc. Sleep usually works in cycles and repeats itself by transitioning into different stages of sleep. This study is unique in that it uses wearable devices to collect multiple parameters from subjects and uses this information to predict sleep stages and sleep patterns. For the multivariate multiclass sleep stage prediction problem, we have experimented with both memoryless (ML) and memory-based models on seven database instances, that is, five from the collected dataset and two from the existing datasets. The Random Forest classifier outclassed the ML models that are LR, MLP, kNN, and SVM with accuracy (ACC) of 0.96 and Cohen Kappa 0.96, and the memory-based model long short-term memory (LSTM) performed well on all the datasets with the maximum attained accuracy of 0.88 and Kappa 0.82. The proposed methodology was also validated on a longitudinal dataset, the Multiethnic Study of Atherosclerosis (MESA), with ACC and Kappa of 0.75 and 0.64 for ML models and 0.86 and 0.78 for memory-based models, respectively, and from another benchmarked Apple Watch dataset available on Physio-Net with ACC and Kappa of 0.93 and 0.93 for ML and 0.92 and 0.87 for memory-based models, respectively. The given methodology showed better results than the original work and indicates that the memory-based method works better to capture the sleep pattern.

Distinct lateral hypothalamic CaMKIIα neuronal populations regulate wakefulness and locomotor activity

For nearly a century, evidence has accumulated indicating that the lateral hypothalamus (LH) contains neurons essential to sustain wakefulness. While lesion or inactivation of LH neurons produces a profound increase in sleep, stimulation of inhibitory LH neurons promotes wakefulness. To date, the primary wake-promoting cells that have been identified in the LH are the hypocretin/orexin (Hcrt) neurons, yet these neurons have little impact on total sleep or wake duration across the 24-h period. Recently, we and others have identified other LH populations that increase wakefulness. In the present study, we conducted microendoscopic calcium imaging in the LH concomitant with EEG and locomotor activity (LMA) recordings and found that a subset of LH neurons that express Ca2+/calmodulin-dependent protein kinase IIα (CaMKIIα) are preferentially active during wakefulness. Chemogenetic activation of these neurons induced sustained wakefulness and greatly increased LMA even in the absence of Hcrt signaling. Few LH CaMKIIα-expressing neurons are hypocretinergic or histaminergic while a small but significant proportion are GABAergic. Ablation of LH inhibitory neurons followed by activation of the remaining LH CaMKIIα neurons induced similar levels of wakefulness but blunted the LMA increase. Ablated animals showed no significant changes in sleep architecture but both spontaneous LMA and high theta (8 to 10 Hz) power during wakefulness were reduced. Together, these findings indicate the existence of two subpopulations of LH CaMKIIα neurons: an inhibitory population that promotes locomotion without affecting sleep architecture and an excitatory population that promotes prolonged wakefulness even in the absence of Hcrt signaling.

Sleep-Disordered Breathing Destabilizes Ventricular Repolarization

Rationale

Sleep-disordered breathing (SDB) increases the risk of cardiac arrhythmias and sudden cardiac death.

Objectives

To characterize the associations between SDB, intermittent hypoxemia, and the beat-to-beat QT variability index (QTVI), a measure of ventricular repolarization lability associated with a higher risk for cardiac arrhythmias, sudden cardiac death, and mortality.

Methods

Three distinct cohorts were used for the current study. The first cohort, used for cross-sectional analysis, was a matched sample of 122 participants with and without severe SDB. The second cohort, used for longitudinal analysis, consisted of a matched sample of 52 participants with and without incident SDB. The cross-sectional and longitudinal cohorts were selected from the Sleep Heart Health Study participants. The third cohort comprised 19 healthy adults exposed to acute intermittent hypoxia and ambient air on two separate days. Electrocardiographic measures were calculated from one-lead electrocardiograms.

Results

Compared to those without SDB, participants with severe SDB had greater QTVI (-1.19 in participants with severe SDB vs. -1.43 in participants without SDB, P = 0.027), heart rate (68.34 vs. 64.92 beats/minute; P = 0.028), and hypoxemia burden during sleep as assessed by the total sleep time with oxygen saturation less than 90% (TST90; 11.39% vs. 1.32%, P < 0.001). TST90, but not the frequency of arousals, was a predictor of QTVI. QTVI during sleep was predictive of all-cause mortality. With incident SDB, mean QTVI increased from -1.23 to -0.86 over 5 years (P = 0.017). Finally, exposing healthy adults to acute intermittent hypoxia for four hours progressively increased QTVI (from -1.85 at baseline to -1.64 after four hours of intermittent hypoxia; P = 0.016).

Conclusions

Prevalent and incident SDB are associated with ventricular repolarization instability, which predisposes to ventricular arrhythmias and sudden cardiac death. Intermittent hypoxemia destabilizes ventricular repolarization and may contribute to increased mortality in SDB.

Differences in sleep spindle wave density between patients with diabetes mellitus and matched controls: implications for sensing and regulation of peripheral blood glucose

Background

Brain waves during sleep are involved in sensing and regulating peripheral glucose level. Whether brain waves in patients with diabetes differ from those of healthy subjects is unknown. We examined the hypothesis that patients with diabetes have reduced sleep spindle waves, a form of brain wave implicated in periphery glucose regulation during sleep.

Methods

From a retrospective analysis of polysomnography (PSG) studies on patients who underwent sleep apnea evaluation, we identified 1,214 studies of patients with diabetes mellitus (>66% type 2) and included a sex- and age-matched control subject for each within the scope of our analysis. We similarly identified 376 patients with prediabetes and their matched controls. We extracted spindle characteristics from artifact-removed PSG electroencephalograms and other patient data from records. We used rank-based statistical methods to test hypotheses. We validated our finding on an external PSG dataset.

Results

Patients with diabetes mellitus exhibited on average about half the spindle density (median=0.38 spindles/min) during sleep as their matched control subjects (median=0.70 spindles/min) (P<2.2e-16). Compared to controls, spindle loss was more pronounced in female patients than in male patients in the frontal regions of the brain (P=0.04). Patients with prediabetes also exhibited signs of lower spindle density compared to matched controls (P=0.01-0.04).

Conclusions

Patients with diabetes have fewer spindle waves that are implicated in glucose regulation than matched controls during sleep. Besides offering a possible explanation for neurological complications from diabetes, our findings open the possibility that reversing/reducing spindle loss could improve the overall health of patients with diabetes mellitus.

Non-Invasive Biosensing for Healthcare Using Artificial Intelligence: A Semi-Systematic Review

The rapid development of biosensing technologies together with the advent of deep learning has marked an era in healthcare and biomedical research where widespread devices like smartphones, smartwatches, and health-specific technologies have the potential to facilitate remote and accessible diagnosis, monitoring, and adaptive therapy in a naturalistic environment. This systematic review focuses on the impact of combining multiple biosensing techniques with deep learning algorithms and the application of these models to healthcare. We explore the key areas that researchers and engineers must consider when developing a deep learning model for biosensing: the data modality, the model architecture, and the real-world use case for the model. We also discuss key ongoing challenges and potential future directions for research in this field. We aim to provide useful insights for researchers who seek to use intelligent biosensing to advance precision healthcare.

pyPPG: a Python toolbox for comprehensive photoplethysmography signal analysis

Objective.Photoplethysmography is a non-invasive optical technique that measures changes in blood volume within tissues. It is commonly and being increasingly used for a variety of research and clinical applications to assess vascular dynamics and physiological parameters. Yet, contrary to heart rate variability measures, a field which has seen the development of stable standards and advanced toolboxes and software, no such standards and limited open tools exist for continuous photoplethysmogram (PPG) analysis. Consequently, the primary objective of this research was to identify, standardize, implement and validate key digital PPG biomarkers.Approach.This work describes the creation of a standard Python toolbox, denotedpyPPG, for long-term continuous PPG time-series analysis and demonstrates the detection and computation of a high number of fiducial points and digital biomarkers using a standard fingerbased transmission pulse oximeter.Main results.The improved PPG peak detector had an F1-score of 88.19% for the state-of-the-art benchmark when evaluated on 2054 adult polysomnography recordings totaling over 91 million reference beats. The algorithm outperformed the open-source original Matlab implementation by ∼5% when benchmarked on a subset of 100 randomly selected MESA recordings. More than 3000 fiducial points were manually annotated by two annotators in order to validate the fiducial points detector. The detector consistently demonstrated high performance, with a mean absolute error of less than 10 ms for all fiducial points.Significance.Based on these fiducial points,pyPPGengineered a set of 74 PPG biomarkers. Studying PPG time-series variability usingpyPPGcan enhance our understanding of the manifestations and etiology of diseases. This toolbox can also be used for biomarker engineering in training data-driven models.pyPPGis available onhttps://physiozoo.com/.

Unsupervised ensembling of multiple software sensors with phase synchronization: a robust approach for electrocardiogram-derived respiration

Objective.We aimed to fuse the outputs of different electrocardiogram-derived respiration (EDR) algorithms to create one higher quality EDR signal.Methods.We viewed each EDR algorithm as a software sensor that recorded breathing activity from a different vantage point, identified high-quality software sensors based on the respiratory signal quality index, aligned the highest-quality EDRs with a phase synchronization technique based on the graph connection Laplacian, and finally fused those aligned, high-quality EDRs. We refer to the output as the sync-ensembled EDR signal. The proposed algorithm was evaluated on two large-scale databases of whole-night polysomnograms. We evaluated the performance of the proposed algorithm using three respiratory signals recorded from different hardware sensors, and compared it with other existing EDR algorithms. A sensitivity analysis was carried out for a total of five cases: fusion by taking the mean of EDR signals, and the four cases of EDR signal alignment without and with synchronization and without and with signal quality selection.Results.The sync-ensembled EDR algorithm outperforms existing EDR algorithms when evaluated by the synchronized correlation (γ-score), optimal transport (OT) distance, and estimated average respiratory rate score, all with statistical significance. The sensitivity analysis shows that the signal quality selection and EDR signal alignment are both critical for the performance, both with statistical significance.Conclusion.The sync-ensembled EDR provides robust respiratory information from electrocardiogram.Significance.Phase synchronization is not only theoretically rigorous but also practical to design a robust EDR.

Association of sleep timing with all-cause and cardiovascular mortality: the Sleep Heart Health Study and the Osteoporotic Fractures in Men Study

STUDY OBJECTIVES

Previous studies have highlighted the importance of sleep patterns for human health. This study aimed to investigate the association of sleep timing with all-cause and cardiovascular disease mortality.

METHODS

Participants were screened from two cohort studies: the Sleep Heart Health Study (SHHS; n = 4,824) and the Osteoporotic Fractures in Men Study (n = 2,658). Sleep timing, including bedtime and wake-up time, was obtained from sleep habit questionnaires at baseline. The sleep midpoint was defined as the halfway point between the bedtime and wake-up time. Restricted cubic splines and Cox proportional hazards regression analyses were used to examine the association between sleep timing and mortality.

RESULTS

We observed a U-shaped association between bedtime and all-cause mortality in both the SHHS and Osteoporotic Fractures in Men Study groups. Specifically, bedtime at 11:00 pm and waking up at 7:00 am was the nadir for all-cause and cardiovascular disease mortality risks. Individuals with late bedtime (> 12:00 am) had an increased risk of all-cause mortality in SHHS (hazard ratio 1.53, 95% confidence interval 1.28-1.84) and Osteoporotic Fractures in Men Study (hazard ratio 1.27, 95% confidence interval 1.01-1.58). In the SHHS, late wake-up time (> 8:00 am) was associated with increased all-cause mortality (hazard ratio 1.39, 95% confidence interval 1.13-1.72). No significant association was found between wake-up time and cardiovascular disease mortality. Delaying sleep midpoint (> 4:00 am) was also significantly associated with all-cause mortality in the SHHS and Osteoporotic Fractures in Men Study.

CONCLUSIONS

Sleep timing is associated with all-cause and cardiovascular disease mortality. Our findings highlight the importance of appropriate sleep timing in reducing mortality risk.

CITATION

Ma M, Fan Y, Peng Y, et al. Association of sleep timing with all-cause and cardiovascular mortality: the Sleep Heart Health Study and the Osteoporotic Fractures in Men Study. J Clin Sleep Med. 2024;20(4):545-553.

Research and application of deep learning-based sleep staging: Data, modeling, validation, and clinical practice

Over the past few decades, researchers have attempted to simplify and accelerate the process of sleep stage classification through various approaches; however, only a few such approaches have gained widespread acceptance. Artificial intelligence technology, particularly deep learning, is promising for earning the trust of the sleep medicine community in automated sleep-staging systems, thus facilitating its application in clinical practice and integration into daily life. We aimed to comprehensively review the latest methods that are applying deep learning for enhancing sleep staging efficiency and accuracy. Starting from the requisite "data" for constructing deep learning algorithms, we elucidated the current landscape of this domain and summarized the fundamental modeling process, encompassing signal selection, data pre-processing, model architecture, classification tasks, and performance metrics. Furthermore, we reviewed the applications of automated sleep staging in scenarios such as sleep-disorder screening, diagnostic procedures, and health monitoring and management. Finally, we conducted an in-depth analysis and discussion of the challenges and future in intelligent sleep staging, particularly focusing on large-scale sleep datasets, interdisciplinary collaborations, and human-computer interactions.

Alpha anteriorization and theta posteriorization during deep sleep

Brain states (wake, sleep, general anesthesia, etc.) are profoundly associated with the spatiotemporal dynamics of brain oscillations. Previous studies showed that the EEG alpha power shifted from the occipital cortex to the frontal cortex (alpha anteriorization) after being induced into a state of general anesthesia via propofol. The sleep research literature suggests that slow waves and sleep spindles are generated locally and propagated gradually to different brain regions. Since sleep and general anesthesia are conceptualized under the same framework of consciousness, the present study examines whether alpha anteriorization similarly occurs during sleep and how the EEG power in other frequency bands changes during different sleep stages. The results from the analysis of three polysomnography datasets of 234 participants show consistent alpha anteriorization during the sleep stages N2 and N3, beta anteriorization during stage REM, and theta posteriorization during stages N2 and N3. Although it is known that the neural circuits responsible for sleep are not exactly the same for general anesthesia, the findings of alpha anteriorization in this study suggest that, at macro level, the circuits for alpha oscillations are organized in the similar cortical areas. The spatial shifts of EEG power in different frequency bands during sleep may offer meaningful neurophysiological markers for the level of consciousness.

Integrating Phenotypic Information of Obstructive Sleep Apnea and Deep Representation of Sleep-Event Sequences for Cardiovascular Risk Prediction

Background

Advances in mobile, wearable and machine learning (ML) technologies for gathering and analyzing long-term health data have opened up new possibilities for predicting and preventing cardiovascular diseases (CVDs). Meanwhile, the association between obstructive sleep apnea (OSA) and CV risk has been well-recognized. This study seeks to explore effective strategies of incorporating OSA phenotypic information and overnight physiological information for precise CV risk prediction in the general population.

Methods

1,874 participants without a history of CVDs from the MESA dataset were included for the 5-year CV risk prediction. Four OSA phenotypes were first identified by the K-mean clustering based on static polysomnographic (PSG) features. Then several phenotype-agnostic and phenotype-specific ML models, along with deep learning (DL) models that integrate deep representations of overnight sleep-event feature sequences, were built for CV risk prediction. Finally, feature importance analysis was conducted by calculating SHapley Additive exPlanations (SHAP) values for all features across the four phenotypes to provide model interpretability.

Results

All ML models showed improved performance after incorporating the OSA phenotypic information. The DL model trained with the proposed phenotype-contrastive training strategy performed the best, achieving an area under the Receiver Operating Characteristic (ROC) curve of 0.877. Moreover, PSG and FOOD FREQUENCY features were recognized as significant CV risk factors across all phenotypes, with each phenotype emphasizing unique features.

Conclusion

Models that are aware of OSA phenotypes are preferred, and lifestyle factors should be a greater focus for precise CV prevention and risk management in the general population.

Objective sleep characteristics and hypertension: a community-based cohort study

Objective

The link between sleep quality and hypertension risk is well-established. However, research on the specific dose-relationship between objective sleep characteristics and hypertension incidence remains limited. This study aims to explore the dose-relationship association between objective sleep characteristics and hypertension incidence.

Methods

A community-based prospective cohort study design was employed using data from the Sleep Heart Health Study (SHHS). A total of 2,460 individuals were included in the study, of which 780 had hypertension. Baseline personal characteristics and medical history were collected. Objective sleep characteristics were obtained through polysomnography (PSG). Multivariate logistic regression models were utilized for analysis. Restricted cubic splines (RCS) were used to examine dose-relationship associations.

Results

After adjusting for covariates, the percentage of total sleep duration in stage 2 (N2%) was positively associated with hypertension incidence, while the N3% was negatively associated with hypertension incidence Odds ratio (OR) = 1.009, 95% confidence interval (CI) [1.001, 1.018], P = 0.037; OR = 0.987, 95% CI: [0.979, 0.995], P = 0.028, respectively. For every 10% increase in N2 sleep, the risk of developing hypertension increases by 9%, while a 3% decrease in N3 sleep corresponds to a 0.1% increase in the incidence of hypertension. In the subgroup of non-depression, a positive association between N2% and hypertension was significant statistically (OR = 1.012, 95%CI, 1.002, 1.021, P = 0.013, Pinteraction = 0.013). RCS demonstrated that the risk of developing hypertension was lower when N2% ranged from 38% to 58% and rapidly increased thereafter (P = 0.002, non-linear P = 0.040). The lowest risk for hypertension incidence risk of N3% occurring at 25%, and a significant increase below 15% or above 40% (P = 0.001, non-linear P = 0.008).

Conclusions

There's a negative association between N3% and the incidence of hypertension, and a positive association between N2% and the incidence of hypertension, particularly among non-depression individuals. These associations exhibit strong non-linear dose-response relationships.

Fatigued but not sleepy? An empirical investigation of the differentiation between fatigue and sleepiness in sleep disorder patients in a cross-sectional study

OBJECTIVE

Sleepiness and fatigue are common complaints among individuals with sleep disorders. The two concepts are often used interchangeably, causing difficulty with differential diagnosis and treatment decisions. The current study investigated sleep disorder patients to determine which factors best differentiated sleepiness from fatigue.

METHODS

The study used a subset of participants from a multi-site study (n = 606), using a cross-sectional study design. We selected 60 variables associated with either sleepiness or fatigue, including demographic, mental health, and lifestyle factors, medical history, sleep questionnaires, rest-activity rhythms (actigraphy), polysomnographic (PSG) variables, and sleep diaries. Fatigue was measured with the Fatigue Severity Scale and sleepiness was measured with the Epworth Sleepiness Scale. A Random Forest machine learning approach was utilized for analysis.

RESULTS

Participants' average age was 47.5 years (SD 14.0), 54.6% female, and the most common sleep disorder diagnosis was obstructive sleep apnea (67.4%). Sleepiness and fatigue were moderately correlated (r = 0.334). The model for fatigue (explained variance 49.5%) indicated depression was the strongest predictor (relative explained variance 42.7%), followed by insomnia severity (12.3%). The model for sleepiness (explained variance 17.9%), indicated insomnia symptoms was the strongest predictor (relative explained variance 17.6%). A post hoc receiver operating characteristic analysis indicated depression could be used to discriminate fatigue (AUC = 0.856) but not sleepiness (AUC = 0.643).

CONCLUSIONS

The moderate correlation between fatigue and sleepiness supports previous literature that the two concepts are overlapping yet distinct. Importantly, depression played a more prominent role in characterizing fatigue than sleepiness, suggesting depression could be used to differentiate the two concepts.

Is the time below 90% of SpO2 during sleep (T90%) a metric of good health? A longitudinal analysis of two cohorts

BACKGROUND

Novel wireless-based technologies can easily record pulse oximetry at home. One of the main parameters that are recorded in sleep studies is the time under 90% of SpO2 (T90%) and the oxygen desaturation index 3% (ODI-3%). We assessed the association of T90% and/or ODI-3% in two different scenarios (a community-based study and a clinical setting) with all-cause mortality (primary outcome).

METHODS

We included all individuals from the Sleep Heart Health Study (SHHS, community-based cohort) and Santiago Obstructive Sleep Apnea (SantOSA, clinical cohort) with complete data at baseline and follow-up. Two measures of hypoxemia (T90% and ODI-3%) were our primary exposures. The adjusted hazard ratios (HRs) per standard deviation (pSD) between T90% and incident all-cause mortality (primary outcome) were determined by adjusted Cox regression models. In the secondary analysis, to assess whether T90% varies across clinical factors, anthropometrics, abdominal obesity, metabolic rate, and SpO2, we conducted linear regression models. Incremental changes in R2 were conducted to test the hypothesis.

RESULTS

A total of 4323 (56% male, median 64 years old, follow-up: 12 years, 23% events) and 1345 (77% male, median 55 years old, follow-up: 6 years, 11.6% events) patients were included in SHHS and SantOSA, respectively. Every 1 SD increase in T90% was associated with an adjusted HR of 1.18 [95% CI: 1.10-1.26] (p value < 0.001) in SHHS and HR 1.34 [95% CI: 1.04-1.71] (p value = 0.021) for all-cause mortality in SantOSA. Conversely, ODI-3% was not associated with worse outcomes. R2 explains 62% of the variability in T90%. The main contributors were baseline-mean change in SpO2, baseline SpO2, respiratory events, and age.

CONCLUSION

The findings suggest that T90% may be an important marker of wellness in clinical and community-based scenarios. Although this nonspecific metric varies across the populations, ventilatory changes during sleep rather than other physiological or comorbidity variables explain their variability.

Dealing with missing data in multi-informant studies: A comparison of approaches

Multi-informant studies are popular in social and behavioral science. However, their data analyses are challenging because data from different informants carry both shared and unique information and are often incomplete. Using Monte Carlo Simulation, the current study compares three approaches that can be used to analyze incomplete multi-informant data when there is a distinction between reference and nonreference informants. These approaches include a two-method measurement model for planned missing data (2MM-PMD), treating nonreference informants' reports as auxiliary variables with the full-information maximum likelihood method or multiple imputation, and listwise deletion. The result suggests that 2MM-PMD, when correctly specified and data are missing at random, has the best overall performance among the examined approaches regarding point estimates, type I error rates, and statistical power. In addition, it is also more robust to data that are not missing at random.

Individualized temporal patterns dominate cortical upstate and sleep depth in driving human sleep spindle timing

Sleep spindles are critical for memory consolidation and strongly linked to neurological disease and aging. Despite their significance, the relative influences of factors like sleep depth, cortical up/down states, and spindle temporal patterns on individual spindle production remain poorly understood. Moreover, spindle temporal patterns are typically ignored in favor of an average spindle rate. Here, we analyze spindle dynamics in 1008 participants from the Multi-Ethnic Study of Atherosclerosis using a point process framework. Results reveal fingerprint-like temporal patterns, characterized by a refractory period followed by a period of increased spindle activity, which are highly individualized yet consistent night-to-night. We observe increased timing variability with age and distinct gender/age differences. Strikingly, and in contrast to the prevailing notion, individualized spindle patterns are the dominant determinant of spindle timing, accounting for over 70% of the statistical deviance explained by all of the factors we assessed, surpassing the contribution of slow oscillation (SO) phase (~14%) and sleep depth (~16%). Furthermore, we show spindle/SO coupling dynamics with sleep depth are preserved across age, with a global negative shift towards the SO rising slope. These findings offer novel mechanistic insights into spindle dynamics with direct experimental implications and applications to individualized electroencephalography biomarker identification.

A meta-analysis of the association between insomnia with objective short sleep duration and risk of hypertension

The aim of this meta-analysis was to examine the association between insomnia with objective short sleep duration (ISSD) with prevalent and incident hypertension in cross-sectional and longitudinal studies, respectively. Data were collected from 6 cross-sectional studies with 5914 participants and 2 longitudinal studies with 1963 participants. Odds ratios (ORs) for prevalent and risk ratios (RRs) for incident hypertension were calculated through meta-analyses of adjusted data from individual studies. Compared to normal sleepers with objective normal sleep duration (NNSD), ISSD was significantly associated with higher pooled OR for prevalent hypertension (pooled OR = 2.67, 95%CI = 1.45-4.90) and pooled RR for incident hypertension (pooled RR = 1.95, 95%CI = 1.19-3.20), respectively. Compared to insomnia with objective normal sleep duration, ISSD was associated with significantly higher pooled OR of prevalent hypertension (pooled OR = 1.94, 95%CI = 1.29-2.92) and pooled RR for incident hypertension (pooled RR = 2.07, 95%CI = 1.47-2.90), respectively. Furthermore, normal sleepers with objective short sleep duration were not associated with either prevalent (pooled OR = 1.21, 95%CI = 0.84-1.75) or incident (pooled RR = 0.97, 95%CI = 0.81-1.17) hypertension compared to NNSD. Our findings suggest that ISSD is a more severe phenotype of the disorder associated with a higher risk of hypertension. Objective short sleep duration might be a valid and clinically useful index of insomnia's impact on cardiovascular health.

Major Depressive Disorder Prediction Based on Sleep-Wake Disorders Symptoms in US Adolescents: A Machine Learning Approach from National Sleep Research Resource

Background

There is substantial evidence from previous studies that abnormalities in sleep parameters associated with depression are demonstrated in almost all stages of sleep architecture. Patients with symptoms of sleep-wake disorders have a much higher risk of developing major depressive disorders (MDD) compared to those without.

Objective

The aim of the present study is to establish and compare the performance of different machine learning models based on sleep-wake disorder symptoms data and to select the optimal model to interpret the importance of sleep-wake disorder symptoms to predict MDD occurrence in adolescents.

Methods

We derived data for this work from 2020 to 2021 Assessing Nocturnal Sleep/Wake Effects on Risk of Suicide Phase I Study from National Sleep Research Resource. Using demographic and sleep-wake disorder symptoms data as predictors and the occurrence of MDD measured base on the center for epidemiologic studies depression scale as an outcome, the following six machine learning predictive models were developed: eXtreme Gradient Boosting model (XGBoost), Light Gradient Boosting mode, AdaBoost, Gaussian Naïve Bayes, Complement Naïve Bayes, and multilayer perceptron. The models' performance was assessed using the AUC and other metrics, and the final model's predictor importance ranking was explained.

Results

XGBoost is the optimal predictive model in comprehensive performance with the AUC of 0.804 in the test set. All sleep-wake disorder symptoms were significantly positively correlated with the occurrence of adolescent MDD. The insomnia severity was the most important predictor compared with the other predictors in this study.

Conclusion

This machine learning predictive model based on sleep-wake disorder symptoms can help to raise the awareness of risk of symptoms between sleep-wake disorders and MDD in adolescents and improve primary care and prevention.

InsightSleepNet: the interpretable and uncertainty-aware deep learning network for sleep staging using continuous Photoplethysmography

BACKGROUND

This study was conducted to address the existing drawbacks of inconvenience and high costs associated with sleep monitoring. In this research, we performed sleep staging using continuous photoplethysmography (PPG) signals for sleep monitoring with wearable devices. Furthermore, our aim was to develop a more efficient sleep monitoring method by considering both the interpretability and uncertainty of the model's prediction results, with the goal of providing support to medical professionals in their decision-making process.

METHOD

The developed 4-class sleep staging model based on continuous PPG data incorporates several key components: a local attention module, an InceptionTime module, a time-distributed dense layer, a temporal convolutional network (TCN), and a 1D convolutional network (CNN). This model prioritizes both interpretability and uncertainty estimation in its prediction results. The local attention module is introduced to provide insights into the impact of each epoch within the continuous PPG data. It achieves this by leveraging the TCN structure. To quantify the uncertainty of prediction results and facilitate selective predictions, an energy score estimation is employed. By enhancing both the performance and interpretability of the model and taking into consideration the reliability of its predictions, we developed the InsightSleepNet for accurate sleep staging.

RESULT

InsightSleepNet was evaluated using three distinct datasets: MESA, CFS, and CAP. Initially, we assessed the model's classification performance both before and after applying an energy score threshold. We observed a significant improvement in the model's performance with the implementation of the energy score threshold. On the MESA dataset, prior to applying the energy score threshold, the accuracy was 84.2% with a Cohen's kappa of 0.742 and weighted F1 score of 0.842. After implementing the energy score threshold, the accuracy increased to a range of 84.8-86.1%, Cohen's kappa values ranged from 0.75 to 0.78 and weighted F1 scores ranged from 0.848 to 0.861. In the case of the CFS dataset, we also noted enhanced performance. Before the application of the energy score threshold, the accuracy stood at 80.6% with a Cohen's kappa of 0.72 and weighted F1 score of 0.808. After thresholding, the accuracy improved to a range of 81.9-85.6%, Cohen's kappa values ranged from 0.74 to 0.79 and weighted F1 scores ranged from 0.821 to 0.857. Similarly, on the CAP dataset, the initial accuracy was 80.6%, accompanied by a Cohen's kappa of 0.73 and weighted F1 score was 0.805. Following the application of the threshold, the accuracy increased to a range of 81.4-84.3%, Cohen's kappa values ranged from 0.74 to 0.79 and weighted F1 scores ranged from 0.813 to 0.842. Additionally, by interpreting the model's predictions, we obtained results indicating a correlation between the peak of the PPG signal and sleep stage classification.

CONCLUSION

InsightSleepNet is a 4-class sleep staging model that utilizes continuous PPG data, serves the purpose of continuous sleep monitoring with wearable devices. Beyond its primary function, it might facilitate in-depth sleep analysis by medical professionals and empower them with interpretability for intervention-based predictions. This capability can also support well-informed clinical decision-making, providing valuable insights and serving as a reliable second opinion in medical settings.

Loop Gain as a Predictor of Blood Pressure Response in Patients Treated for Obstructive Sleep Apnea: Secondary Analysis of a Clinical Trial

Rationale: Randomized trials have shown inconsistent cardiovascular benefits from obstructive sleep apnea (OSA) therapy. Intermittent hypoxemia can increase both sympathetic nerve activity and loop gain ("ventilatory instability"), which may thus herald cardiovascular treatment benefit. Objectives: To test the hypothesis that loop gain predicts changes in 24-hour mean blood pressure (MBP) in response to OSA therapy and compare its predictive value against that of other novel biomarkers. Methods: The HeartBEAT (Heart Biomarker Evaluation in Apnea Treatment) trial assessed the effect of 12 weeks of continuous positive airway pressure (CPAP) versus oxygen versus control on 24-hour MBP. We measured loop gain and hypoxic burden from sleep tests and identified subjects with a sleepy phenotype using cluster analysis. Associations between biomarkers and 24-h MBP were assessed in the CPAP/oxygen arms using linear regression models adjusting for various covariates. Secondary outcomes and predictors were analyzed similarly. Results: We included 93 and 94 participants in the CPAP and oxygen arms, respectively. Overall, changes in 24-hour MBP were small, but interindividual variability was substantial (mean [standard deviation], -2 [8] and 1 [8] mm Hg in the CPAP and oxygen arms, respectively). Higher loop gain was significantly associated with greater reductions in 24-hour MBP independent of covariates in the CPAP arm (-1.5 to -1.9 mm Hg per 1-standard-deviation increase in loop gain; P ⩽ 0.03) but not in the oxygen arm. Other biomarkers were not associated with improved cardiovascular outcomes. Conclusions: To our knowledge, this is the first study suggesting that loop gain predicts blood pressure response to CPAP therapy. Eventually, loop gain estimates may facilitate patient selection for research and clinical practice. Clinical trial registered with www.clinicaltrials.gov (NCT01086800).

Automatic Sleep Staging Based on Contextual Scalograms and Attention Convolution Neural Network Using Single-Channel EEG

Single-channel EEG based sleep staging is of interest to researchers due to its broad application prospect in daily sleep monitoring recently. We proposed using contextual scalograms as input and developed a convolutional neural network with attention modules named Co-ScaleNet for sleep staging. The contextual scalograms were obtained by combining the same color channels of three original RGB scalograms from consecutive epochs, and a simple and efficient data augmentation was designed according to their various forms. The Co-ScaleNet consists of two main parts. Firstly, three parallel convolutional branches with attention modules correspondingly extract and fuse features from contextual scalograms at the top layers. The remaining part is a stack of lightweight blocks. We achieved an overall accuracy of 87.0% for healthy individuals, 84.7% for depressed patients. And we obtained comparable performance on the public Sleep-EDFx (82.8%), ISRUC (84.6%) and SHHS datasets (87.7%), including a high recall of N1. The contextual scalograms of R channel as input achieved the best performance, which conform to the features of interest in visual scoring. The attention modules improved the recall of N1 and N3. Overall, the contextual scalograms provided a novel scheme for both contextual information extraction and data augmentation. Our study successfully expanded its application to depression datasets, as well as patients with sleep apnea, demonstrating its wide applicability.

Nocturnal oxygen saturation is associated with all-cause mortality: a community-based study

STUDY OBJECTIVES

Observational studies have demonstrated the association between the single-point measurement of oxygen saturation (SpO2) level and mortality in the general population. This study aimed to evaluate whether nocturnal SpO2 level could predict all-cause mortality in a community-based population.

METHODS

The study samples were obtained from the Sleep Heart Health Study, which included 2,280 men and 2,606 women (mean age, 63.8 ± 11.1 years). A pulse oximeter based on overnight in-home polysomnography was used to monitor SpO2 levels during total sleep time (SpO2-TOTAL). Multivariable Cox proportional hazards analysis was performed to examine the association between nocturnal SpO2 and all-cause mortality.

RESULTS

During the follow-up period of 10.7 ± 3.0 years, 1,110 (22.7%) people died. After adjusting for confounding factors, multivariable Cox regression analysis showed that the average SpO2-TOTAL (hazard ratio [HR] 0.93; 95% confidence interval [CI] 0.90-0.96, P < .001) was associated with all-cause mortality. These findings remained stable in individuals with low and high apnea-hypopnea index levels. Additionally, maximum SpO2-TOTAL (HR, 0.91; 95% CI, 0.87-0.96; P = .001) and minimum SpO2-TOTAL (HR, 0.98; 95% CI, 0.97-0.99; P = .001) could predict all-cause mortality. A significant association between nocturnal hypoxemia and all-cause mortality was also observed.

CONCLUSIONS

Our findings highlight the importance of monitoring nocturnal SpO2 level and improving hypoxemia in the general populations.

CITATION

Yan B, Gao Y, Zhang Z, Shi T, Chen Q. Nocturnal oxygen saturation is associated with all-cause mortality: a community-based study. J Clin Sleep Med. 2024;20(2):229-235.

Assessing Risk of Health Outcomes From Brain Activity in Sleep: A Retrospective Cohort Study

Background and Objectives

Patterns of electrical activity in the brain (EEG) during sleep are sensitive to various health conditions even at subclinical stages. The objective of this study was to estimate sleep EEG-predicted incidence of future neurologic, cardiovascular, psychiatric, and mortality outcomes.

Methods

This is a retrospective cohort study with 2 data sets. The Massachusetts General Hospital (MGH) sleep data set is a clinic-based cohort, used for model development. The Sleep Heart Health Study (SHHS) is a community-based cohort, used as the external validation cohort. Exposure is good, average, or poor sleep defined by quartiles of sleep EEG-predicted risk. The outcomes include ischemic stroke, intracranial hemorrhage, mild cognitive impairment, dementia, atrial fibrillation, myocardial infarction, type 2 diabetes, hypertension, bipolar disorder, depression, and mortality. Diagnoses were based on diagnosis codes, brain imaging reports, medications, cognitive scores, and hospital records. We used the Cox survival model with death as the competing risk.

Results

There were 8673 participants from MGH and 5650 from SHHS. For all outcomes, the model-predicted 10-year risk was within the 95% confidence interval of the ground truth, indicating good prediction performance. When comparing participants with poor, average, and good sleep, except for atrial fibrillation, all other 10-year risk ratios were significant. The model-predicted 10-year risk ratio closely matched the observed event rate in the external validation cohort.

Discussion

The incidence of health outcomes can be predicted by brain activity during sleep. The findings strengthen the concept of sleep as an accessible biological window into unfavorable brain and general health outcomes.

MixSleepNet: A Multi-Type Convolution Combined Sleep Stage Classification Model

BACKGROUND AND OBJECTIVE

Sleep staging is an essential step for sleep disorder diagnosis, which is time-intensive and laborious for experts to perform this work manually. Automatic sleep stage classification methods not only alleviate experts from these demanding tasks but also enhance the accuracy and efficiency of the classification process.

METHODS

A novel multi-channel biosignal-based model constructed by the combination of a 3D convolutional operation and a graph convolutional operation is proposed for the automated sleep stages using various physiological signals. Both the 3D convolution and graph convolution can aggregate information from neighboring brain areas, which helps to learn intrinsic connections from the biosignals. Electroencephalogram (EEG), electromyogram (EMG), electrooculogram (EOG) and electrocardiogram (ECG) signals are employed to extract time domain and frequency domain features. Subsequently, these signals are input to the 3D convolutional and graph convolutional branches, respectively. The 3D convolution branch can explore the correlations between multi-channel signals and multi-band waves in each channel in the time series, while the graph convolution branch can explore the connections between each channel and each frequency band. In this work, we have developed the proposed multi-channel convolution combined sleep stage classification model (MixSleepNet) using ISRUC datasets (Subgroup 3 and 50 random samples from Subgroup 1).

RESULTS

Based on the first expert's label, our generated MixSleepNet yielded an accuracy, F1-score and Cohen kappa scores of 0.830, 0.821 and 0.782, respectively for ISRUC-S3. It obtained accuracy, F1-score and Cohen kappa scores of 0.812, 0.786, and 0.756, respectively for the ISRUC-S1 dataset. In accordance with the evaluations conducted by the second expert, the comprehensive accuracies, F1-scores, and Cohen kappa coefficients for the ISRUC-S3 and ISRUC-S1 datasets are determined to be 0.837, 0.820, 0.789, and 0.829, 0.791, 0.775, respectively.

CONCLUSION

The results of the performance metrics by the proposed method are much better than those from all the compared models. Additional experiments were carried out on the ISRUC-S3 sub-dataset to evaluate the contributions of each module towards the classification performance.

Stability and Volatility of Human Rest-Activity Rhythms: Insights from Very Long Actograms (VLAs)

Importance

Wrist-worn activity monitors provide biomarkers of health by non-obtrusively measuring the timing and amount of rest and physical activity (rest-activity rhythms, RARs). The morphology and robustness of RARs vary by age, gender, and sociodemographic factors, and are perturbed in various chronic illnesses. However, these are cross-sectionally derived associations from recordings lasting 4-10 days, providing little insights into how RARs vary with time.

Objective

To describe how RAR parameters can vary or evolve with time (~months).

Design Setting and Participants

48 very long actograms ("VLAs", ≥90 days in duration) were identified from subjects enrolled in the STAGES (Stanford Technology, Analytics and Genomics in Sleep) study, a prospective cross-sectional, multi-site assessment of individuals > 13 years of age that required diagnostic polysomnography to address a sleep complaint. A single 3-year long VLA (author GD) is also described.

Exposures/Intervention

None planned.

Main Outcomes and Measures

For each VLA, we assessed the following parameters in 14-day windows: circadian/ultradian spectrum, pseudo-F statistic ("F"), cosinor amplitude, intradaily variability, interdaily stability, acrophase and estimates of "sleep" and non-wearing.

Results

Included STAGES subjects (n = 48, 30 female) had a median age of 51, BMI of 29.4kg/m2, Epworth Sleepiness Scale score (ESS) of 10/24 and a median recording duration of 120 days. We observed marked within-subject undulations in all six RAR parameters, with many subjects displaying ultradian rhythms of activity that waxed and waned in intensity. When appraised at the group level (nomothetic), averaged RAR parameters remained remarkably stable over a ~4 month recording period. Cohort-level deficits in average RAR robustness associated with unemployment or high BMI (>29.4) also remained stable over time.

Conclusions and Relevance

Through an exemplary set of months-long wrist actigraphy recordings, this study quantitatively depicts the longitudinal stability and dynamic range of human rest-activity rhythms. We propose that continuous and long-term actigraphy may have broad potential as a holistic, transdiagnostic and ecologically valid monitoring biomarker of changes in chronobiological health. Prospective recordings from willing subjects will be necessary to precisely define contexts of use.

Deep transfer learning for automated single-lead EEG sleep staging with channel and population mismatches

Introduction: Automated sleep staging using deep learning models typically requires training on hundreds of sleep recordings, and pre-training on public databases is therefore common practice. However, suboptimal sleep stage performance may occur from mismatches between source and target datasets, such as differences in population characteristics (e.g., an unrepresented sleep disorder) or sensors (e.g., alternative channel locations for wearable EEG). Methods: We investigated three strategies for training an automated single-channel EEG sleep stager: pre-training (i.e., training on the original source dataset), training-from-scratch (i.e., training on the new target dataset), and fine-tuning (i.e., training on the original source dataset, fine-tuning on the new target dataset). As source dataset, we used the F3-M2 channel of healthy subjects (N = 94). Performance of the different training strategies was evaluated using Cohen's Kappa (κ) in eight smaller target datasets consisting of healthy subjects (N = 60), patients with obstructive sleep apnea (OSA, N = 60), insomnia (N = 60), and REM sleep behavioral disorder (RBD, N = 22), combined with two EEG channels, F3-M2 and F3-F4. Results: No differences in performance between the training strategies was observed in the age-matched F3-M2 datasets, with an average performance across strategies of κ = .83 in healthy, κ = .77 in insomnia, and κ = .74 in OSA subjects. However, in the RBD set, where data availability was limited, fine-tuning was the preferred method (κ = .67), with an average increase in κ of .15 to pre-training and training-from-scratch. In the presence of channel mismatches, targeted training is required, either through training-from-scratch or fine-tuning, increasing performance with κ = .17 on average. Discussion: We found that, when channel and/or population mismatches cause suboptimal sleep staging performance, a fine-tuning approach can yield similar to superior performance compared to building a model from scratch, while requiring a smaller sample size. In contrast to insomnia and OSA, RBD data contains characteristics, either inherent to the pathology or age-related, which apparently demand targeted training.

A robust deep learning detector for sleep spindles and K-complexes: towards population norms

Sleep spindles (SSs) and K-complexes (KCs) are brain patterns involved in cognitive functions that appear during sleep. Large-scale sleep studies would benefit from precise and robust automatic sleep event detectors, capable of adapting the variability in both electroencephalography (EEG) signals and expert annotation rules. We introduce the Sleep EEG Event Detector (SEED), a deep learning system that outperforms existing approaches in SS and KC detection, reaching an F1-score of 80.5% and 83.7%, respectively, on the MASS2 dataset. SEED transfers well and requires minimal fine-tuning for new datasets and annotation styles. Remarkably, SEED substantially reduces the required amount of annotated data by using a novel pretraining approach that leverages the rule-based detector A7. An analysis of 11,224 subjects revealed that SEED's detections provide better estimates of SS population statistics than existing approaches. SEED is a powerful resource for obtaining sleep-event statistics that could be useful for establishing population norms.