New and Emerging Approaches to Better Define Sleep Disruption and Its Consequences

Strengths, weaknesses, opportunities, and threats of using AI-enabled technology in sleep medicine: a commentary

Over the past few years, artificial intelligence (AI) has emerged as a powerful tool used to efficiently automate several tasks across multiple domains. Sleep medicine is perfectly positioned to leverage this tool due to the wealth of physiological signals obtained through sleep studies or sleep tracking devices and abundance of accessible clinical data through electronic medical records. However, caution must be applied when utilizing AI, due to intrinsic challenges associated with novel technology. The Artificial Intelligence in Sleep Medicine Committee of the American Academy of Sleep Medicine reviews advancements in AI within the sleep medicine field. In this article, the Artificial Intelligence in Sleep Medicine committee members provide a commentary on the scope of AI technology in sleep medicine. The commentary identifies 3 pivotal areas in sleep medicine that can benefit from AI technologies: clinical care, lifestyle management, and population health management. This article provides a detailed analysis of the strengths, weaknesses, opportunities, and threats associated with using AI-enabled technologies in each pivotal area. Finally, the article broadly reviews barriers and challenges associated with using AI-enabled technologies and offers possible solutions.

CITATION

Bandyopadhyay A, Oks M, Sun H, et al. Strengths, weaknesses, opportunities, and threats of using AI-enabled technology in sleep medicine: a commentary. J Clin Sleep Med. 2024;20(7):1183-1191.

Reconsidering screening thresholds in health assessments for obstructive sleep apnea using operational and safety incident data

The rail industry in Australia screens workers for probable obstructive sleep apnea (OSA) due to known safety risks. However, existing criteria to trigger screening only identify a small proportion of workers with OSA. The current study sought to examine the relationship between OSA risk and rail incidents in real-world data from Australian train drivers, and conducted a proof of concept analysis to determine whether more conservative screening criteria are justified. Health assessment (2016-2018) and subsequent rail incident data (2016-2020) were collected from two passenger rail service providers. Predictors included OSA status (confirmed no OSA with a sleep study, controlled OSA, unknown OSA [no recorded sleep assessment data] and confirmed OSA with no indication of treatment); OSA risk according to the current Standard, and OSA risk according to more conservative clinical markers (BMI threshold and cardiometabolic burden). Coded rail safety incidents involving the train driver were included. Data were analysed using zero-inflated negative binomial models to account for over-dispersion with high 0 counts, and rail safety incidents are reported using Incidence Risk Ratios (IRRs). A total of 751 train drivers, typically middle-aged, overweight to obese and mostly men, were included in analyses. There were 43 (5.7%) drivers with confirmed OSA, 62 (8.2%) with controlled OSA, 13 (1.7%) with confirmed no OSA and 633 (84.4%) drivers with unknown OSA. Of the 633 train drivers with unknown OSA status, 21 (3.3%) met 'at risk' criteria for OSA according to the Standard, and incidents were 61% greater (IRR: 1.61, 95% Confidence Interval (CI) 1.02-2.56) in the years following their health assessment compared to drivers who did not meet 'at risk' criteria. A more conservative OSA risk status using lower BMI threshold and cardiometabolic burden identified an additional 30 'at risk' train drivers who had 46% greater incidents compared to drivers who did not meet risk criteria (IRR (95% CI) 1.46 (1.00-2.13)). Our more conservative OSA risk criteria identified more workers, with greater prospective incidents. These findings suggest that existing validated tools could be considered in future iterations of the Standard in order to more sensitively screen for OSA.

Automated sleep staging on reduced channels in children with epilepsy

Objectives

This study aimed to validate a sleep staging algorithm using in-hospital video-electroencephalogram (EEG) in children without epilepsy, with well-controlled epilepsy (WCE), and with drug-resistant epilepsy (DRE).

Methods

Overnight video-EEG, along with electrooculogram (EOG) and chin electromyogram (EMG), was recorded in children between 4 and 18 years of age. Classical sleep staging was performed manually as a ground truth. An end-to-end hierarchical recurrent neural network for sequence-to-sequence automatic sleep staging (SeqSleepNet) was used to perform automated sleep staging using three channels: C4-A1, EOG, and chin EMG.

Results

In 176 children sleep stages were manually scored: 47 children without epilepsy, 74 with WCE, and 55 with DRE. The 5-class sleep staging accuracy of the automatic sleep staging algorithm was 84.7% for the children without epilepsy, 83.5% for those with WCE, and 80.8% for those with DRE (Kappa of 0.79, 0.77, and 0.73 respectively). Performance per sleep stage was assessed with an F1 score of 0.91 for wake, 0.50 for N1, 0.83 for N2, 0.84 for N3, and 0.86 for rapid eye movement (REM) sleep.

Conclusion

We concluded that the tested algorithm has a high accuracy in children without epilepsy and with WCE. Performance in children with DRE was acceptable, but significantly lower, which could be explained by a tendency of more time spent in N1, and by abundant interictal epileptiform discharges and intellectual disability leading to less recognizable sleep stages. REM sleep time, however, significantly affected in children with DRE, can be detected reliably by the algorithm.Clinical trial registration: ClinicalTrials.gov, identifier NCT04584385.

Causal Association Between Subtypes of Excessive Daytime Sleepiness and Risk of Cardiovascular Diseases

BACKGROUND

Excessive daytime sleepiness (EDS), experienced in 10% to 20% of the population, has been associated with cardiovascular disease and death. However, the condition is heterogeneous and is prevalent in individuals having short and long sleep duration. We sought to clarify the relationship between sleep duration subtypes of EDS with cardiovascular outcomes, accounting for these subtypes.

METHODS AND RESULTS

We defined 3 sleep duration subtypes of excessive daytime sleepiness: normal (6-9 hours), short (<6 hours), and long (>9 hours), and compared these with a nonsleepy, normal-sleep-duration reference group. We analyzed their associations with incident myocardial infarction (MI) and stroke using medical records of 355 901 UK Biobank participants and performed 2-sample Mendelian randomization for each outcome. Compared with healthy sleep, long-sleep EDS was associated with an 83% increased rate of MI (hazard ratio, 1.83 [95% CI, 1.21-2.77]) during 8.2-year median follow-up, adjusting for multiple health and sociodemographic factors. Mendelian randomization analysis provided supporting evidence of a causal role for a genetic long-sleep EDS subtype in MI (inverse-variance weighted β=1.995, P=0.001). In contrast, we did not find evidence that other subtypes of EDS were associated with incident MI or any associations with stroke (P>0.05).

CONCLUSIONS

Our study suggests the previous evidence linking EDS with increased cardiovascular disease risk may be primarily driven by the effect of its long-sleep subtype on higher risk of MI. Underlying mechanisms remain to be investigated but may involve sleep irregularity and circadian disruption, suggesting a need for novel interventions in this population.

Multi-night measurement for diagnosis and simplified monitoring of obstructive sleep apnoea

Substantial night-to-night variability in obstructive sleep apnoea (OSA) severity has raised misdiagnosis and misdirected treatment concerns with the current prevailing single-night diagnostic approach. In-home, multi-night sleep monitoring technology may provide a feasible complimentary diagnostic pathway to improve both the speed and accuracy of OSA diagnosis and monitor treatment efficacy. This review describes the latest evidence on night-to-night variability in OSA severity, and its impact on OSA diagnostic misclassification. Emerging evidence for the potential impact of night-to-night variability in OSA severity to influence important health risk outcomes associated with OSA is considered. This review also characterises emerging diagnostic applications of wearable and non-wearable technologies that may provide an alternative, or complimentary, approach to traditional OSA diagnostic pathways. The required evidence to translate these devices into clinical care is also discussed. Appropriately sized randomised controlled trials are needed to determine the most appropriate and effective technologies for OSA diagnosis, as well as the optimal number of nights needed for accurate diagnosis and management. Potential risks versus benefits, patient perspectives, and cost-effectiveness of these novel approaches should be carefully considered in future trials.

Creating an Optimal Approach for Diagnosing Sleep Apnea

Sleep apnea is nowadays recognized as a treatable chronic disease and awareness of it has increased, leading to an upsurge in demand for diagnostic testing. Conventionally, diagnosis depends on overnight polysomnography in a sleep clinic, which is highly human-resource intensive and ignores the night-to-night variability in classical sleep apnea markers, such as the apnea-hypopnea index. In this review, the authors summarize the main improvements that could be made in the sleep apnea diagnosis strategy; how technological innovations and multi-night home testing could be used to simplify, increase access, and reduce costs of diagnostic testing while avoiding misclassification of severity.

Morphic Sensors for Respiratory Parameters Estimation: Validation against Overnight Polysomnography

Effective monitoring of respiratory disturbances during sleep requires a sensor capable of accurately capturing chest movements or airflow displacement. Gold-standard monitoring of sleep and breathing through polysomnography achieves this task through dedicated chest/abdomen bands, thermistors, and nasal flow sensors, and more detailed physiology, evaluations via a nasal mask, pneumotachograph, and airway pressure sensors. However, these measurement approaches can be invasive and time-consuming to perform and analyze. This work compares the performance of a non-invasive wearable stretchable morphic sensor, which does not require direct skin contact, embedded in a t-shirt worn by 32 volunteer participants (26 males, 6 females) with sleep-disordered breathing who performed a detailed, overnight in-laboratory sleep study. Direct comparison of computed respiratory parameters from morphic sensors versus traditional polysomnography had approximately 95% (95 ± 0.7) accuracy. These findings confirm that novel wearable morphic sensors provide a viable alternative to non-invasively and simultaneously capture respiratory rate and chest and abdominal motions.

High night-to-night variability in sleep apnea severity is associated with uncontrolled hypertension

Obstructive sleep apnea (OSA) severity can vary markedly from night-to-night. However, the impact of night-to-night variability in OSA severity on key cardiovascular outcomes such as hypertension is unknown. Thus, the primary aim of this study is to determine the effects of night-to-night variability in OSA severity on hypertension likelihood. This study uses in-home monitoring of 15,526 adults with ~180 nights per participant with an under-mattress sleep sensor device, plus ~30 repeat blood pressure measures. OSA severity is defined from the mean estimated apnea-hypopnoea index (AHI) over the ~6-month recording period for each participant. Night-to-night variability in severity is determined from the standard deviation of the estimated AHI across recording nights. Uncontrolled hypertension is defined as mean systolic blood pressure ≥140 mmHg and/or mean diastolic blood pressure ≥90 mmHg. Regression analyses are performed adjusted for age, sex, and body mass index. A total of 12,287 participants (12% female) are included in the analyses. Participants in the highest night-to-night variability quartile within each OSA severity category, have a 50-70% increase in uncontrolled hypertension likelihood versus the lowest variability quartile, independent of OSA severity. This study demonstrates that high night-to-night variability in OSA severity is a predictor of uncontrolled hypertension, independent of OSA severity. These findings have important implications for the identification of which OSA patients are most at risk of cardiovascular harm.

Single-Night Diagnosis of Sleep Apnea Contributes to Inconsistent Cardiovascular Outcome Findings

BACKGROUND

Single-night disease misclassification of OSA due to night-to-night variability may contribute to inconsistent findings in OSA trials.

RESEARCH QUESTION

Does multinight quantification of OSA severity provide more precise estimates of associations with incident hypertension?

STUDY DESIGN AND METHODS

A total of 3,831 participants without hypertension at baseline were included in simulation analyses. Included participants had ≥ 28 days of nightly apnea-hypopnea index (AHI) recordings via an under-mattress sensor and ≥ 3 separate BP measurements over a 3-month baseline period followed by ≥ 3 separate BP measurements 6 to 9 months postbaseline. Incident hypertension was defined as a mean systolic BP ≥ 140 mm Hg or a mean diastolic BP ≥ 90 mm Hg. Simulated trials (1,000) were performed, using bootstrap methods to investigate the effect of variable numbers of nights (x = 1-56 per participant) to quantify AHI and the ability to detect associations between OSA and incident hypertension via logistic regression adjusted for age, sex, and BMI.

RESULTS

Participants were middle-aged (mean ± SD, 52 ± 12 y), mostly men (91%), and overweight (BMI, 28 ± 5 kg/m²). Single-night quantification of OSA failed to detect an association with hypertension risk in 42% of simulated trials (α = 0.05). Conversely, 100% of trials detected an association when AHI was quantified over ≥ 28 nights. Point estimates of hypertension risk were also 50% higher and uncertainty was 5 times lower during multinight vs single-night simulation trials.

INTERPRETATION

Multinight monitoring of OSA allows for better estimates of hypertension risk and potentially other adverse health outcomes associated with OSA. These findings have important implications for clinical care and OSA trial design.

An experimental investigation on the impact of wind turbine noise on polysomnography-measured and sleep diary-determined sleep outcomes

Study Objectives

Carefully controlled studies of wind turbine noise (WTN) and sleep are lacking, despite anecdotal complaints from some residents in wind farm areas and known detrimental effects of other noises on sleep. This laboratory-based study investigated the impact of overnight WTN exposure on objective and self-reported sleep outcomes.

Methods

Sixty-eight participants (38 females) aged (mean ± SD) 49.2 ± 19.5 were recruited from four groups; N = 14, living &lt;10 km from a wind farm and reporting WTN related sleep disruption; N = 18, living &lt;10 km from a wind farm and reporting no WTN sleep disruption; N = 18, reporting road traffic noise-related sleep disruption; and N = 18 control participants living in a quiet rural area. All participants underwent in-laboratory polysomnography during four full-night noise exposure conditions in random order: a quiet control night (19 dB(A) background laboratory noise), continuous WTN (25 dB(A)) throughout the night; WTN (25 dB(A)) only during periods of established sleep; and WTN (25 dB(A)) only during periods of wake or light N1 sleep. Group, noise condition, and interaction effects on measures of sleep quantity and quality were examined via linear mixed model analyses.

Results

There were no significant noise condition or group-by-noise condition interaction effects on polysomnographic or sleep diary determined sleep outcomes (all ps &gt; .05).

Conclusions

These results do not support that WTN at 25 dB(A) impacts sleep outcomes in participants with or without prior WTN exposure or self-reported habitual noise-related sleep disruption. These findings do not rule out effects at higher noise exposure levels or potential effects of WTN on more sensitive markers of sleep disruption.

Clinical Trial Registration

ACTRN12619000501145, UTN U1111-1229-6126. Establishing the physiological and sleep disruption characteristics of noise disturbances in sleep. https://www.anzctr.org.au/. This study was prospectively registered on the Australian and New Zealand Clinical Trial Registry.

A novel EEG marker predicts perceived sleepiness and poor sleep quality

STUDY OBJECTIVES

To determine if a novel EEG-derived continuous index of sleep depth/alertness, the odds ratio product (ORP), predicts self-reported daytime sleepiness and poor sleep quality in two large population-based cohorts.

METHODS

ORP values which range from 0 (deep sleep) to 2.5 (fully alert) were calculated in 3s intervals during awake periods (ORPwake) and NREM sleep (ORPNREM) determined from home sleep studies in the HypnoLaus (N = 2162: 1106 females, 1056 males) and men androgen inflammation lifestyle environment and stress (MAILES) cohorts (N = 754 males). Logistic regression was used to examine associations between ORPwake, ORPNREM, and traditional polysomnography measures (as comparators) with excessive sleepiness (Epworth sleepiness scale >10) and poor sleep quality (Pittsburgh sleep quality index >5) and insomnia symptoms.

RESULTS

High ORPwake was associated with a ~30% increase in poor sleep quality in both HypnoLaus (odds ratio, OR, and 95% CI) 1.28 (1.09, 1.51), and MAILES 1.36 (1.10, 1.68). High ORPwake was also associated with a ~28% decrease in excessive daytime sleepiness in the MAILES dataset. ORPNREM was associated with a ~30% increase in poor sleep quality in HypnoLaus but not in MAILES. No consistent associations across cohorts were detected using traditional polysomnography markers.

CONCLUSIONS

ORP, a novel EEG-derived metric, measured during wake periods predicts poor sleep quality in two independent cohorts. Consistent with insomnia symptomatology of poor perceived sleep in the absence of excessive daytime sleepiness, ORPwake may provide valuable objective mechanistic insight into physiological hyperarousal.

All-Cause Mortality in People with Co-Occurring Insomnia Symptoms and Sleep Apnea: Analysis of the Wisconsin Sleep Cohort

PURPOSE

Insomnia symptoms and sleep apnea frequently co-occur and are associated with worse sleep, daytime function, mental health and quality of life, compared to either insomnia or obstructive sleep apnea (OSA) alone. This study aimed to investigate the association of symptoms of co-morbid insomnia and sleep apnea (COMISA) with all-cause mortality.

PATIENTS AND METHODS

Wisconsin Sleep Cohort data were analysed to assess potential associations between COMISA symptoms and all-cause mortality. Nocturnal insomnia symptoms were defined as difficulties initiating sleep, maintaining sleep, and/or early morning awakenings "often" or "almost always", and/or regular sedative-hypnotic medicine use. OSA was defined as an apnea-hypopnea index ≥5/hr sleep. Participants were classified as having neither insomnia symptoms nor OSA, insomnia symptoms alone, OSA alone, or COMISA symptoms. Associations between the four groups and all-cause mortality over 20 years of follow-up were examined via multivariable adjusted Cox regression models.

RESULTS

Among 1115 adult participants (mean ± SD age 55 ± 8 years, 53% males), 19.1% had COMISA symptoms. After controlling for sociodemographic and behavioral factors, COMISA symptoms were associated with an increased risk of all-cause mortality compared to no insomnia symptoms or OSA (HR [95% CI]; 1.71 [1.00-2.93]). OSA alone (0.91 [0.53, 1.57]) and insomnia symptoms alone (1.04 [0.55, 1.97]) were not associated with increased mortality risk.

CONCLUSION

Co-morbid insomnia symptoms and sleep apnea is associated with increased all-cause mortality risk. Future research should investigate mechanisms underpinning COMISA and the effectiveness of different treatment approaches to reduce mortality risk for this common condition.

Environmental noise-induced cardiovascular responses during sleep

STUDY OBJECTIVES

This study was designed to test the utility of cardiovascular responses as markers of potentially different environmental noise disruption effects of wind farm compared to traffic noise exposure during sleep.

METHODS

Twenty participants underwent polysomnography. In random order, and at six sound pressure levels from 33 dBA to 48 dBA in 3 dB increments, three types of wind farm and two types of road traffic noise recordings of 20-sec duration were played during established N2 or deeper sleep, each separated by 20 seconds without noise. Each noise sequence also included a no-noise control. Electrocardiogram and finger pulse oximeter recorded pulse wave amplitude changes from the pre-noise onset baseline following each noise exposure and were assessed algorithmically to quantify the magnitude of heart rate and finger vasoconstriction responses to noise exposure.

RESULTS

Higher sound pressure levels were more likely to induce drops in pulse wave amplitude. Sound pressure levels as low as 39 dBA evoked a pulse wave amplitude response (Odds ratio [95% confidence interval]; 1.52 [1.15, 2.02]). Wind farm noise with amplitude modulation was less likely to evoke a pulse wave amplitude response than the other noise types, but warrants cautious interpretation given low numbers of replications within each noise type.

CONCLUSION

These preliminary data support that drops in pulse wave amplitude are a particularly sensitive marker of noise-induced cardiovascular responses during. Larger trials are clearly warranted to further assess relationships between recurrent cardiovascular activation responses to environmental noise and potential long-term health effects.

Multi-night Prevalence, Variability, and Diagnostic Misclassification of Obstructive Sleep Apnea

Rationale Recent studies suggest that obstructive sleep apnea (OSA) severity can vary markedly from night to night which may have important implications for diagnosis and management. Objectives This study aimed to assess OSA prevalence from multi-night in-home recordings and the impact of night-to-night variability in OSA severity on diagnostic classification in a large, global, non-randomly selected community sample from a consumer database of people that purchased a novel, validated, under-mattress sleep analyzer. Methods 67,278 individuals aged between 18 and 90 years underwent in-home nightly monitoring over an average of ~170 nights per participant between July 2020 to March 2021. OSA was defined as a nightly mean apnea-hypopnea index (AHI) >15 events/h. Outcomes were multi-night global prevalence and likelihood of OSA misclassification from a single night AHI value. Measurements and Main Results Over 11.6 million nights of data were collected and analyzed. OSA global prevalence was 22.6% (95% CI: 20.9-24.3%). The likelihood of misdiagnosis in people with OSA based on a single night ranged between ~20% and 50%. Misdiagnosis error rates decreased with increased monitoring nights (e.g. 1-night F1-score=0.77 vs. 0.94 for 14-nights); and remained stable after 14-nights of monitoring. Conclusions Multi-night in-home monitoring using novel non-invasive under mattress sensor technology indicates a global prevalence of moderate to severe OSA of ~20%, and that ~20% of people diagnosed with a single night study may be misclassified. These findings highlight the need to consider night-to-night variation on OSA diagnosis and management. This article is open access and distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives License 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Co-morbid insomnia and obstructive sleep apnoea is associated with all-cause mortality

STUDY OBJECTIVES

Increased mortality has been reported in people with insomnia and in those with obstructive sleep apnoea (OSA). However, these conditions commonly co-occur and the combined effect of co-morbid insomnia and sleep apnoea (COMISA) on mortality risk is unknown. This study used Sleep Heart Health Study (SHHS) data to assess associations between COMISA and all-cause mortality risk.

METHODS

Insomnia was defined as difficulties falling asleep, maintaining sleep, and/or early morning awakenings from sleep ≥16 times a month and daytime impairment. OSA was defined as an apnoea-hypopnoea index ≥15 events/h sleep. COMISA was defined if both conditions were present. Multivariable adjusted Cox proportional hazard models were used to determine the association between COMISA and all-cause mortality (n=1210) over 15 years of follow-up.

RESULTS

5236 participants were included. 2708 (52%) did not have insomnia/OSA (control), 170 (3%) had insomnia-alone, 2221 (42%) had OSA-alone, and 137 (3%) had COMISA. COMISA participants had a higher prevalence of hypertension (ORs [95%CI]; 2.00 [1.39, 2.90]) and cardiovascular disease compared to controls (1.70 [1.11, 2.61]). Insomnia-alone and OSA-alone were associated with higher risk of hypertension but not cardiovascular disease compared to controls. Compared to controls, COMISA was associated with a 47% (HR, 95% CI; 1.47 (1.06, 2.07)) increased risk of mortality. The association between COMISA and mortality was consistent across multiple definitions of OSA and insomnia.

CONCLUSIONS

Co-morbid insomnia and sleep apnoea was associated with higher rates of hypertension and cardiovascular disease at baseline, and an increased risk of all-cause mortality compared to no insomnia/OSA.