EEG & Sleep Analysis

Pat & Dennis Bender Early Dementia Diagnosis & Prognosis Fund

 

J. Dennis Bender

Office, Home & Cell Phone: 859-391-5226

5726 La Jolla Blvd. – Suite 311

La Jolla, CA 92037-7345

&

Office - 100 Riverside Pl. - Suite 303

Covington, KY 41011-5711

 

We support the development of improved diagnostic methods for the early detection and diagnosis of MCI, Alzheimer’s, vascular and other dementias, their likely prognosis, and best treatment options. We focus on the development of Bayesian-based, medical-decision-support systems, comparative-effectiveness research, and the better utilization of these for the above. (After incorporating in KY as a 501(c)3 in 2002, we dissolved that entity in favor of a simplified form of two entirely self-financed, private philanthropies utilizing a Vanguard Charitable Trust for making annual-research-grants for early-dementia-detection and its correct differential-diagnosis and likely-prognosis. They will continue on, after I am long gone, either mentally or physically, with annual grants. Scripps Foundation, Profs. Randall Bateman, James Brewer and others will be our fund’s future research grant advisors. KMK Law is our legal advisor and my estate executor is Elizabeth Dunn.

 (See: https://www.alz.org/alzheimers-dementia/research_progress/earlier-diagnosis)

 

 This email address is being protected from spambots. You need JavaScript enabled to view it.

www.JDBender.com – EMS/eVTOL & Educational Experimental Aviation Fund (Vanguard Charitable Trust)

www.JDBender.org – Dementia Diagnosis Fund (Vanguard Charitable Trust)

 

December 27, 2024

 

[Derek Brahney]

“Your sleep-tracker might give you information about more than just your sleep -- specifically, it might give you information about chronic conditions. It's not just whether you sleep well or not -- it's the patterns-of-sleep-over-time where the key info hides. . . Detect-early-disease-biomarkers, personalize-therapies, and monitor-neurological-health and treatment-response.”

 

Here is my very-long, 75+page discussion investigating this important topic related to the early-detection and diagnosis of MCI and dementia of all types. I spent the holiday-week digging deeply into the subject to decide which diagnostic-device to purchase next and how much to invest in this sort of research in the coming year.

 

UCSD-based researchers found that little changes in sleep-quality helped them identify health-risks using data from an Oura-Ring. Those little changes wouldn't show up on an average-night, or on a questionnaire, so it shows how wearables could help us detect risks that would otherwise be missed.

 

Better yet, EEG-signal-analysis from headbands, in addition to various sleep and other measures, might be a useful addition to our collection of early-dementia-detection bio-signals. Examples such as MUSE-2 or -S: Headband Meditation Tracker and Monitor and even better the DREEM-3: https://beacon.bio/dreem-headband - Professional model used in an earlier comparison with other commercially available versions.

 

The DREEM-3 is my current favorite and is available from Beacon Biosignals, a health technology company transforming the development of treatments for neurological, psychiatric, and sleep disorders. With advanced machine-learning, wearable-EEG-devices, large real-world neurophysiology-datasets, and multiple-FDA-clearances, Beacon provides insights into brain-activity to improve patient outcomes [https://beacon.bio.] However, it is a more-expensive alternative and requires a subscription to use it, so there is a tradeoff. The comparison research paper at the end discusses all of this in great detail, so that you can choose the best alternative to fit your own needs, versus my own.

 

Statement of Significance: The development of ambulatory-technologies able to monitor physiological-signals during sleep at-home and longitudinally is rising. These technologies advancements have the potential to move forward the field of sleep-medicine, but the poor validation of current wearable-technology inhibits their widespread-use. This validation study of a reduced-montage dry-electroencephalographic (EEG) device showed that this device is able to acquire EEG, heart-rate, and breathing-frequency and automatically analyze these signals using machine-learning approach to provide sleep-stages with an accuracy close to the consensus of 5 professional sleep-scorers.

 

Original Article: (April 23, 2020) The Dreem-Headband compared to polysomnography for electroencephalographic signal acquisition and sleep staging

 

Pierrick J. Arnal1, *, Valentin Thorey2 , Eden Debellemaniere1 , Michael E. Ballard1 , Albert Bou Hernandez2 , Antoine Guillot2 , Hugo Jourde2 , Mason Harris2 , Mathias Guillard3,4 , Pascal Van Beers3,4 , Mounir Chennaoui3,4 and Fabien Sauvet3,4

 

1 Dreem, Science Team, New York, NY, 2 Dreem, Algorithm Team, Paris, France, 3 French Armed Forces Biomedical Research Institute (IRBA), Fatigue and Vigilance Unit, Bretigny-sur-Orge, France, 4 EA 7330 VIFASOM, Paris Descartes University, Paris, France

 

*Corresponding author. Pierrick J. Arnal, Dreem, Science Team, 450 Park Ave S, New York, NY 10016. Email: This email address is being protected from spambots. You need JavaScript enabled to view it..

 

Abstract Study Objectives: The development of ambulatory technologies capable of monitoring brain activity during sleep longitudinally is critical for advancing sleep science. The aim of this study was to assess the signal-acquisition and the performance of the automatic sleep staging algorithms of a reduced-montage dry-electroencephalographic (EEG) device (Dreem-headband, DH) compared to the gold-standard polysomnography (PSG) scored by 5 sleep experts.

 

Methods: A total of 25 subjects who completed an overnight sleep study at a sleep center while wearing both a PSG and the DH simultaneously have been included in the analysis. We assessed (1) similarity of measured EEG-brain-waves between the DH and the PSG; (2) the heart-rate, breathing-frequency, and respiration-rate-variability (RRV) agreement between the DH and the PSG; and (3) the performance of the DH’s automatic-sleep-staging according to American Academy of Sleep Medicine guidelines versus PSG sleep experts manual-scoring.

 

Results: The mean-percentage-error between the EEG signals acquired by the DH and those from the PSG for the monitoring of α was 15 ± 3.5%, 16 ± 4.3% for β, 16 ± 6.1% for λ, and 10 ± 1.4% for θ frequencies during sleep. The mean-absolute-error for heart-rate, breathing-frequency, and RRV was 1.2 ± 0.5 bpm, 0.3 ± 0.2 cpm, and 3.2 ± 0.6%, respectively.

Automatic-sleep-staging reached an overall accuracy of 83.5 ± 6.4% (F1-score: 83.8 ± 6.3) for the DH to be compared with an average of 86.4 ± 8.0% (F1-score: 86.3 ± 7.4) for the 5 sleep experts.

 

Conclusions: These results demonstrate the capacity of the DH to both monitor sleep-related physiological signals and process them accurately into sleep-stages. This device paves the way for, large-scale, longitudinal-sleep-studies.

 

Clinical Trial Registration: NCT03725943.

 

Key words: sleep; EEG; machine-learning; sleep-stages; device

 

DH Algorithm The DH embedded automatic algorithm works in two stages: (1) feature extraction and (2) classification. It is able to provide real-time sleep staging predictions. (1) Feature extraction is performed for each new epoch of 30 s. Features extracted from the various sensors are concatenated to go through the classification layer. EEG features include power frequency in the λ, α, θ, and β bands and ratio of relative powers as described in 23. Sleep patterns (e.g. slow-oscillations, α-rhythm, spindles, and K-complexes) are detected using an expert approach.

 

The accelerometer provides breathing, movement, and position features. The pulse-oximeter provides cardiac-features. A total of 79 features are extracted from each raw-DH-record. (2) The classification module is built from two layers of Long-Short-Term-Memory [33] and a Softmax-function outputting the final probability-prediction that the epoch belongs to each sleep-stage. It relies on the features extracted from the last 30-epochs to predict the current one. Hence, it takes into account the past temporal-context to make a prediction, as a sleep-expert would do. This classification-module is trained using backpropagation. The training has been done on a dataset composed of previously recorded internal Dreem records.

 

A total of 423 records were used for training and presented several times to the network. A total of 213 validation records from other subjects were used to stop the training when the performance-metrics computed on this validation-set were not improving anymore. None of the records of the current study were used to train or validate the network. We used the framework provided by Pytorch [34] and trained on a single Nvidia Titan X GPU (∼1 h of training, ∼1 s for inference).

 

 

 

Figure 1. 20-s samples of raw-signals recorded by DH (pink) and PSG (black) on the same record during each sleep stages (N1, N2, N3, REM, and Wake). The derivations are F7–O1 for the DH and F3–O1 for the PSG. The signals are presented between −150 and 150 μV.

 

 

Figure 2. Relative spectral power (α, β, λ, and θ frequencies, AU), heart-rate (beats-per-minute, BPM), breathing-frequency (cycles-per-minute, CPM), and respiratory-rate-variability (RRV, %) for a representative-record (i.e. with a MPE similar to the mean of the group). These signals are presented for a whole record for both the DH (pink) and PSG (black).

 

DH Automated-Algorithmic-Approach Finally, we showed that the DH was able to perform real-time sleep-staging using data collected by the DH with an accuracy in the range of individual scorers using PSG data and comparable to the accuracy between PSG scorers in other studies [8, 31].

 

To our knowledge, this performance on a dry-EEG-wearable has never been achieved with another device. Sleep-variables are macro-metrics computed on the hypnogram and are less-impacted than sleep-staging-metrics by local-differences. For instance, wake is slightly underestimated but that does not significantly impact sleep-variables related to wake (WASO, sleep-latency, and sleep-efficiency). Even though the inter-scorer-reliability achieved with PSG by our 5 scorers was high, it highlights the need for such validation studies to rely on a consensus of multiple sleep experts when analyzing sleep staging performance [31].

 

Mixing sleep-experts from different sleep-centers provides a more-realistic analysis than is typically obtained in a standard clinical sleep study where records are scored by only a single individual, which strengthens our results. To evaluate these individual-scorers, we introduced an objective methodology to build a consensus from the other scorers. This enables a fair evaluation of both individual-scorers and the automated-algorithmic-approach of the DH.

 

The main limitation of this study is that the sample was somewhat small and homogeneous in age and sleeper-profile; even though this is consistent with the majority of similar validation studies [20, 22, 23]. A larger sample of more diverse sleepers would have provided more reliability and generalizability to the general population. Therefore, further studies should be run with this investigational device on specific sub-populations (e.g. patients with sleep apnea, psychiatric or neurodegenerative diseases, etc.) either to confirm the performance of the automatic sleep stage algorithm but also to evaluate the ease of use of the device and comfort in these specific targeted populations.

 

In this study, 2.1% of the windows were excluded on average across all the recordings because the virtual-channel could not be computed on the DH-signal due to bad signal-quality on every channel. Since the DH was wore as long as with the PSG, in the lab-setting. It can then be argued that in the home-environment, a more-important-proportion of the signals would have been of bad-quality due to the negligence of the subject. However, it has to be put in line with the fact that in a home-setting, it would have been possible to run multiple nights, which is not easy to do in the lab with a PSG. Also, this issue can be easily coped with an appropriate training of the subject to use the device.

 

Also, this study includes only one night of data per subject with no habituation-night, which may lead to a non-reliable representation of typical sleep in the natural-home-environment, and particularly because sleeping with a full-PSG in a clinical-sleep-lab, which often leads to sleep being shorter and more-fragmented. However, our sample did achieve 87% SE on average, suggesting that sleep was not substantially disrupted on a wide-scale in this study.

 

Conclusion In this study, we showed that using an ambulatory-wireless-dry-EEG device, the DH, it was possible to: (1) acquire EEG-signals that correlate with the EEG-signals recorded with a PSG; (2) reliably measure breathing-frequency and heart-rate continuously during sleep; and (3) perform automatic-sleep-staging-classification according to AASM-criteria with performance similar to that of a consensus of 5 scorers using medical-grade-PSG-data.

 

These results, together with the price, ease-of-use and the availability of raw-signals, pave-the-way for such a device to be an ideal candidate for high-quality, large-scale, longitudinal-sleep-studies in the home or laboratory environment. As such, this technology can enable groundbreaking advancements in sleep-research and medicine. For instance, the resulting database can ultimately be integrated with other types of data collection devices and used to identify unknown patient-subgroups, detect-early-disease-biomarkers, personalize-therapies, and monitor-neurological-health and treatment-response.

 

 

 

 

 

Macro and microarchitectural sleep features in Alzheimer’s dementia and mild cognitive impairment in a large clinical cohort - Jay Pathmanathan, MD PhDAAIC, 2023

 

Search Labs | AI Overview People with Alzheimer's-dementia and mild-cognitive-impairment (MCI) often experience significant sleep-disturbances, including increased-sleep-latency (difficulty falling-asleep), reduced-slow-wave-sleep (deep-sleep), fragmented-sleep, and altered-circadian-rhythms, which can manifest as daytime-sleepiness and disrupted-nighttime-sleep-patterns; these sleep-changes can even precede the onset of noticeable-cognitive-decline and are considered potential-biomarkers for the diseases. 

 

Key sleep features in Alzheimer's and MCI:

• Decreased slow-wave-sleep (SWS): Both Alzheimer's and MCI patients tend to have significantly less deep-sleep (SWS) compared to healthy individuals, impacting memory-consolidation. 
• Increased sleep-latency: Difficulty falling asleep, often taking longer to transition from wakefulness to sleep onset. 
• Sleep-fragmentation: Frequent awakenings during the night, leading to disrupted sleep and reduced sleep efficiency. 
• Altered-REM-sleep: Changes in the duration and timing of rapid-eye-movement (REM) sleep can occur, with some studies showing either reduced or increased REM sleep depending on disease-stage. 
• Daytime-sleepiness: Due to poor nighttime sleep, people with Alzheimer's and MCI may experience excessive daytime sleepiness. 
• Circadian-rhythm-disruption: Irregular-sleep-wake-patterns, with potential for early morning awakenings or napping during the day. 

 

Why sleep-disturbances are important in Alzheimer's and MCI:

• Potential disease-marker: Sleep-disturbances can be an early-indicator of cognitive-decline and may even precede the clinical-diagnosis of Alzheimer's-disease. 
• Impact on cognitive-function: Poor-sleep-quality can negatively affect memory-consolidation, learning, and overall-cognitive-performance. 
• Neurobiological-link: Research suggests that the same brain-pathways involved in sleep-regulation are also affected in Alzheimer's-disease, potentially contributing to the sleep-disturbances observed. 

 

Management of sleep problems in Alzheimer's and MCI:

• Lifestyle-modifications: Maintaining a consistent-sleep-schedule, regular-exercise, creating a relaxing-bedtime-routine, and managing-light-exposure 
• Cognitive-behavioral-therapy (CBT): Addressing underlying thoughts and behaviors that may contribute to sleep-problems 
• Pharmacological-interventions: Depending on the case, medications like melatonin or other sleep-aids may be used to improve sleep-quality 

 

How Sleep Affects AD Pathophysiology [https://pmc.ncbi.nlm.nih.gov/articles/PMC10103826]

It is now well-known that sleep affects cognition. While awake, we make numerous decisions, which result in the formation and accumulation of metabolites. During sleep, there is an increase in the interstitial space in the brain, which leads to the clearance of metabolites from the brain like A-beta lipoprotein [5]. This clearance is done by the glymphatic-system of the brain which is similar to the lymphatic-system of the body, especially during the non-rapid-eye-movement (NREM) SWS-cycle [3]. Clearance leads to decreased fatigue and proper functioning of the brain. In comparison to age-matched-controls, insomnia-patients with lower-SWS had lower-nightly-declarative-memory-consolidation [4].

 

Reduced and fragmented SWS has been linked to greater levels of the amyloid-42 protein in the CSF in older individuals with normal cognitive function [9]. The pathological alterations that underpin AD are thought to begin 10-20 years before any cognitive symptoms occur, with the formation of amyloid plaques in the brain being the earliest detectable preclinical stage of AD [6].

Toxic A-beta lipoprotein is hypothesized to start pathogenic events and lead to the creation of abnormal tau aggregates, which eventually lead to synapse loss and cell death, damaging neural circuitry [4]. Tau is a microtubule-associated-protein found mostly in neuronal-axons, where it helps to maintain microtubule (MIT) structure as well as synaptic-structure and function. The tau-protein is hyperphosphorylated in pathological-states, and it dissociates from MITs, accumulating and producing neurofibrillary-tangles [10].

 

 

Figure 1 portrays the role of sleep in the development of AD.

 

 

The brainstem has been identified to be the earliest brain region to acquire hyperphosphorylated tau, which may be linked to daytime wakefulness instability. Axonal transport is impaired by hyperphosphorylated tau and tangles [10].

 

The orexin (hypocretin) system is affected by A-beta lipoprotein and tau aggregation-mediated neurodegeneration. Orexin is responsible for maintaining alertness, and the lack of orexin is the primary cause of narcolepsy and cataplexy [10]. In studies in mice, both chronic sleep deprivation and orexin infusion raised interstitial fluid A-beta levels, whereas a dual orexin receptor antagonist reduced plaque development [11]. The number of orexin neurons in the hypothalamus is considerably reduced in AD patients in postmortem examination. The absence of or decreased orexin signaling appears to be linked to sleep issues in AD patients, especially in those who sleep excessively throughout the day [10].

 

Amyloid plaques lodge soluble A42, consequently, a decrease in CSF A42 (cerebrospinal fluid (beta)-amyloid 42 protein) indicates the existence of amyloid plaques. One study found that people with low CSF A-beta 42 had poor sleep quality. These people were cognitively intact; therefore, any amyloid accumulation would be regarded as a preclinical stage of AD [6].

 

Toxic A-beta lipoprotein is hypothesized to start pathogenic events and lead to the creation of abnormal tau aggregates, which eventually lead to synapse loss and cell death, damaging neural circuitry [4]. Tau is a microtubule-associated protein found mostly in neuronal axons, where it helps to maintain microtubule (MIT) structure as well as synaptic structure and function. The tau protein is hyperphosphorylated in pathological states, and it dissociates from MITs, accumulating and producing neurofibrillary tangles [10]. Figure 1 portrays the role of sleep in the development of AD.

 

The brainstem has been identified to be the earliest brain region to acquire hyperphosphorylated tau, which may be linked to daytime wakefulness instability. Axonal transport is impaired by hyperphosphorylated tau and tangles [10].

The orexin (hypocretin) system is affected by A-beta lipoprotein and tau aggregation-mediated neurodegeneration. Orexin is responsible for maintaining alertness, and the lack of orexin is the primary cause of narcolepsy and cataplexy [10]. In studies in mice, both chronic sleep deprivation and orexin infusion raised interstitial fluid A-beta levels, whereas a dual orexin receptor antagonist reduced plaque development [11]. The number of orexin neurons in the hypothalamus is considerably reduced in AD patients in postmortem examination. The absence of or decreased orexin signaling appears to be linked to sleep issues in AD patients, especially in those who sleep excessively throughout the day [10].

Amyloid plaques lodge soluble A42, consequently, a decrease in CSF A42 (cerebrospinal fluid (beta)-amyloid 42 protein) indicates the existence of amyloid plaques. One study found that people with low CSF A-beta 42 had poor sleep quality. These people were cognitively intact; therefore, any amyloid accumulation would be regarded as a preclinical stage of AD [6].

Types of Sleep-Disorders in MCI Chronic insomnia has been linked to a slew of negative health outcomes, including a greater risk of high blood pressure, diabetes, obesity, depression, myocardial ischemia, and stroke, as well as a higher chance of cognitive decline and dementia [2]. General sleep alterations observed in healthy older people are increased nocturnal awakenings leading to fragmentation of sleep and decreased daily sleep duration due to equal reductions in both rapid eye movement (REM) and NREM. There is an increase in light sleep (NREM stages 1 and 2) and faster NREM/REM cycles [10]. When compared to age-matched controls, the proportion of SWS, duration of REM sleep, and sleep-wave activity (SWA) in SWS, EEG fall substantially faster in AD patients [10]. REM sleep has desynchronized EEG activity, while NREM sleep has synchronized activity. SWA stands for slow and delta oscillations, as well as isolated slow waves. The high quantities of SWA characterize Stage N3 (NREM Stage 3), also known as delta sleep or SWS. Slow oscillations are generated by the neocortex, and the thalamus helps with their stabilization by temporally altering cortical SWA [4].

One study found that changes in circadian activity rhythms were linked to an increased risk of MCI or dementia in 1,282 older women. Reduced amplitude, a less robust rhythm, and delayed peak activity timing were all implicated in the development of dementia or MCI [11].

A study done on women found that if there was greater variability in sleep efficiency, there was a 1.9-fold higher risk of developing MCI [12]. The above study also found that over the 4.9 years of follow-up, women with the poorest sleep efficiency had 1.5 greater odds of acquiring MCI or dementia, whereas women with longer sleep latencies had 1.4 greater odds of cognitive impairment [12]. Insomnia is characterized by trouble falling asleep, waking up during the night, waking up early and not being able to fall back asleep, and non-restorative sleep [2]. One study after controlling for other demographics/variables/factors showed that for each extra insomnia symptom, there was a 5% increased risk of incident MCI (HR = 1.05; 95% CI: 1.04-1.06) and dementia (HR = 1.05; 95% CI: 1.03-1.06) after 12 years of follow-up [2]. People with poor sleep efficiency spend a longer time in bed to make up for the lost sleep and get the same amount of total sleep time. A comprehensive study that used questionnaires to determine sleep duration indicated that an increase in self-reported sleep duration over time was linked to a two-fold greater risk of cognitive impairment [6]. Women who had a variable total sleep time had a 1.4-fold greater risk of developing MCI compared to women who had less variable total sleep time overall nights [12].

Excessive-daytime-sleepiness (EDS) can be brought on by issues with sleep-hygiene, heart-disease, obesity, drug-use, depression, or sleep-disorders [9]. EDS has previously been linked to diabetes, cardiovascular and cerebrovascular illness, depression, and other conditions that increase the risk of cognitive impairment. According to a study, even after accounting for all of these potentially confounding variables, there is still a substantial correlation between EDS and cognitive-deterioration [13]. Participants with EDS were more-likely to see a drop in MMSE-scores (p = 0.006) [9]. EDS was found to significantly-increase the probability of global-cognitive-decline by 30% while having no significant effect on verbal-fluency or visual-memory-related-activities [13].

How Sleep Helps in Memory Consolidation Memory consolidation occurs during SWS. By coordinating hippocampal-sharp-wave-ripples with thalamic-spindle-activity during slow-oscillation "up" states, cortical-slow-oscillations induce the reactivation of short-term-hippocampal-memories. This mechanism facilitates the consolidation of long-term-memory in the neocortex by contributing to long-term synaptic-plasticity-alterations in neocortical-networks [4]. Sleep-related memory-consolidation was found to be reduced when SWS decreased with age [4]. According to a study, seniors with inconsistent rest-activity-patterns fared worse on composite tests of executive-functioning, memory, and speed than older-persons with more regular rest-activity-patterns [8]. That study also noted that older-adults displayed worsening-alertness, visual-search, response-times, word-identification, addition, anagrams, and object-use after sleep-deprivation than younger-adults [8]. In cognitively-intact-older-people and the initial-stages-of-AD, A-beta-lipoprotein and tau-deposits were linked to lower-NREM-SWA [4]. Researchers discovered evidence of higher-atrophy in the CA2-4-DG region of the hippocampus in insomniacs compared to controls, which was linked to impairments in verbal and visual memory [2].

Interventions Sleep-disorders are common and almost always occur in conjunction with cognitive-decline in patients. The Sleep Study Group of the Italian Dementia Research Association recommends they must always be thoroughly examined using an in-depth-history, physical-examination, and questionnaires and scales, directly written by the patient with the support of the direct caregiver, when possible [13]. Follow-up studies employing objective-sleep/wake-measurements, amyloid-PET-scan, and volumetric-MRI-assessment may be able to track changes in sleep-patterns, brain-amyloid-burden, and neuronal-damage in the elderly with cognitive-impairment [9]. It is generally advisable to start with non-pharmacological-means and then go toward pharmacological-treatments. Some of the interventions are mentioned in Table 1.

Table 1. Pharmacological and non-pharmacological therapies.

TDCS, transcranial direct current stimulation; TMS, transcranial magnetic stimulation

INTERVENTIONS
Non-pharmacological	Physical activity
	Light therapy	Low light, bright light, ambient light
	Mechanical interventions	Rocking beds
	Noninvasive brain stimulation	TDCS, TMS
Pharmacological	Suvorexant, Donepezil

Open in a new tab

Exposure to low-light has been shown to improve-cognitive-function. Bright-light-therapy has also been found to help the elderly enhance and regulate their circadian-activity-rhythm and sleep. Furthermore, ambient-light has been found to affect cognition and affective-mood, and bright-light possesses phase-shifting-qualities that increase-cognitive-performance [11]. Acoustic-stimuli have been proven to improve SWS, and this impacts memory-recall in some patients [10].

Physical-activity has also been demonstrated to have a favorable impact on circadian-activity-rhythms [11]. Rocking-beds have been demonstrated to promote sleep by lowering-sleep-latency, boosting-SWA, and decreasing-nocturnal-arousal. They've also been linked to better memory in people who are otherwise healthy [10].

Restoration of SWA-sleep can be done by noninvasive-brain-stimulation-techniques such as transcranial-direct-current-stimulation (TDCS) and transcranial-magnetic-stimulation (TMS). In a study, TDCS was used to stimulate the brain at the slow-oscillation-frequency in MCI-patients during a daytime-nap; both slow-oscillation-power and memory-performance increased. Moreover, repeated TDCS administrations during SWS caused slow-oscillations and improved-declarative-memory-retention the next-day in both older and younger healthy-people [4].

In one study, insomnia-intervention included sleep-deprivation, cognitive-restructuring, good-sleep-hygiene, exposure to bright-light, alterations in body-temperature, and regular-physical-activity. Following treatment, the treatment-group considerably outperformed the waitlist-group in terms of sleep-onset-latency and sleep-efficiency. In comparison to the waiting-control, treatment was also linked to better performance on complicated-vigilance-tasks [8]. Improvement-in-sleep decreases-oxidative-stress which in turn leads to decreased A-beta-lipoprotein-accumulation [14].

The FDA approved suvorexant, an orexin-receptor antagonist for treating insomnia in AD in the year 2020 [10]. For insomnia, it is recommended that pharmacological therapy be short-term and reviewed regularly every 4-weeks. Long-term-usage of hypnotic-medications should be considered only for specific indications [13].

Pharmaceutical therapies that target the gamma-aminobutyric-acid-system like benzodiazepine and zolpidem, [which I only take very-infrequently,] have been deemed to be less-effective in treating sleep-problems brought on by AD. It has been shown that they affect memory in both people and animals. Additionally, zolpidem use has been linked to an increased chance of dementia [10]. It has been demonstrated that barbiturates and benzodiazepines have detrimental cognitive consequences. Hence, when they are used on the elderly, it is possible that the MCI-symptoms might not be due to neurodegenerative processes but rather these medications [14]. The risk of falls and fractures is also increased by long-term benzodiazepine use.

The cholinergic-system, which encompasses memory and the sleep-wake-cycle, has been linked to the malfunctioning of neurotransmitter-systems involved in sleep. Cholinergic-pathway-neurodegeneration can also disrupt the ascending-reticular-activating-system, resulting in daytime-sleepiness and sleep-disorders [5]. Donepezil, which is an acetylcholinesterase inhibitor, [and which I take daily,] has been proven to increase the amount of REM-sleep in AD-patients [10]. Some people with MCI who are on cholinesterase-inhibitors are frequently prescribed anticholinergics. Dual usage of these two drugs has been found to contribute to functional and cognitive-impairment [14].

Conclusions This paper highlights the pathophysiology of sleep disorders in MCI, different types of sleep disorders, and different interventions that can be used to improve sleep, both pharmacological and non-pharmacological. Based on many articles, there is emerging evidence that sleep affects cognition. Sleep affects our visual and verbal memory, alertness, attention, and performance of simple and complicated tasks. Due to advances in medicine, people are living longer, and so the number of people suffering from cognitive impairment is also increasing. As having a good quality of life is fundamental, it is important to identify sleep disorders as early as possible and intervene. Therefore, there is a need to identify sleep disorders, especially in the elderly, and to investigate them and not just disregard them as a sign of aging. Information like how to maintain good sleep hygiene should be made available to the public. There should also be an awareness of the importance of sleep and its long-term consequences for the public.

The authors have declared that no competing interests exist.

References

1.2019 Alzheimer's disease facts and figures. Alzheimer’s Association. https://alz-journals.onlinelibrary.wiley.com/doi/10.1016/j.jalz.2019.01.010 Alzheimers Dement. 2019;15:321–387. [Google Scholar]

2.Time-varying insomnia symptoms and incidence of cognitive impairment and dementia among older US adults. Resciniti NV, Yelverton V, Kase BE, Zhang J, Lohman MC. Int J Environ Res Public Health. 2021;18:351. doi: 10.3390/ijerph18010351. [DOI] [PMC free article] [PubMed] [Google Scholar]

3.Sleep and dementia. Shenker JI, Singh G. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6140093/ Mo Med. 2017;114:311–315. [PMC free article] [PubMed] [Google Scholar]

4.Slow wave sleep is a promising intervention target for Alzheimer's disease. Lee YF, Gerashchenko D, Timofeev I, Bacskai BJ, Kastanenka KV. Front Neurosci. 2020;14:705. doi: 10.3389/fnins.2020.00705. [DOI] [PMC free article] [PubMed] [Google Scholar]

5.Treatment of sleep disturbance may reduce the risk of future probable Alzheimer's disease. Burke SL, Hu T, Spadola CE, Burgess A, Li T, Cadet T. J Aging Health. 2019;31:322–342. doi: 10.1177/0898264318795567. [DOI] [PMC free article] [PubMed] [Google Scholar]

6.Sleep quality and preclinical Alzheimer disease. Ju YE, McLeland JS, Toedebusch CD, et al. JAMA Neurol. 2013;70:587–593. doi: 10.1001/jamaneurol.2013.2334. [DOI] [PMC free article] [PubMed] [Google Scholar]

7.Association between Z drugs use and risk of cognitive impairment in middle-aged and older patients with chronic insomnia. Guo F, Yi L, Zhang W, Bian ZJ, Zhang YB. Front Hum Neurosci. 2021;15:775144. doi: 10.3389/fnhum.2021.775144. [DOI] [PMC free article] [PubMed] [Google Scholar]

8.Sleep and cognition in older adults. Dzierzewski JM, Dautovich N, Ravyts S. Sleep Med Clin. 2018;13:93–106. doi: 10.1016/j.jsmc.2017.09.009. [DOI] [PMC free article] [PubMed] [Google Scholar]

9.Excessive sleepiness and longer nighttime in bed increase the risk of cognitive decline in frail elderly subjects: The MAPT-sleep study. Gabelle A, Gutierrez LA, Jaussent I, et al. Front Aging Neurosci. 2017;9:312. doi: 10.3389/fnagi.2017.00312. [DOI] [PMC free article] [PubMed] [Google Scholar]

10.Association between sleep, Alzheimer's, and Parkinson's disease. Matsumoto S, Tsunematsu T. Biology (Basel) 2021;10:1127. doi: 10.3390/biology10111127. [DOI] [PMC free article] [PubMed] [Google Scholar]

11.Circadian activity rhythms and risk of incident dementia and mild cognitive impairment in older women. Tranah GJ, Blackwell T, Stone KL, et al. Ann Neurol. 2011;70:722–732. doi: 10.1002/ana.22468. [DOI] [PMC free article] [PubMed] [Google Scholar]

12.Measures of sleep-wake patterns and risk of mild cognitive impairment or dementia in older women. Diem SJ, Blackwell TL, Stone KL, et al. Am J Geriatr Psychiatry. 2016;24:248–258. doi: 10.1016/j.jagp.2015.12.002. [DOI] [PMC free article] [PubMed] [Google Scholar]

13.Sleep disturbances and cognitive decline: recommendations on clinical assessment and the management. Guarnieri B, Cerroni G, Sorbi S. Arch Ital Biol. 2015;153:225–230. doi: 10.12871/0003982920152347. [DOI] [PubMed] [Google Scholar]

14.Potentially inappropriate medication use in older adults with mild cognitive impairment. Weston AL, Weinstein AM, Barton C, Yaffe K. J Gerontol A Biol Sci Med Sci. 2010;65:318–321. doi: 10.1093/gerona/glp158. [DOI] [PMC free article] [PubMed] [Google Scholar]

 

 

 

 

 

Apple Watch Documentation: https://www.apple.com/healthcare/docs/site/Estimating_Sleep_Stages_from_Apple_Watch_Sept_2023.pdf

See Health app [heart-symbol]

 

The Truth About Sleep-Trackers

Their technology is impressive, but imperfect. We asked experts whether they can really help you get a better night’s rest.

By Markham Heid - Nov. 29, 2023 – The New York Times

Shawn McCall, 48, a personal trainer in Waterford, Mich., started tracking his sleep almost 6-years ago. Mr. McCall said that his Oura Ring — a sleek titanium device that he wears on his ring-finger — has revealed how the choices he makes during his waking hours affect his sleep at night.

“It serves as a constant reminder that if I do certain things, like have too much to drink or eat a large meal before bed, I know my heart-rate is going to be higher that night and I’ll definitely have less deep-sleep,” he said. “It helps keep me accountable.”

The popularity of consumer sleep-tracking technology has grown rapidly in recent years, and that growth is projected to continue. Led by wearable devices like the Oura, Fitbit and Apple Watch, the market also includes phone-based apps and “nearables,” which are placed on or beside a person’s bed.

While the capability and sophistication of sleep-trackers varies, they can record things such as heart-rate, movement, body-temperature and blood-oxygen-levels. Using these data, the trackers claim to offer valuable sleep insights, such as estimates of nightly deep-sleep or a total “sleep-score” that reflects overall sleep-quality.

Experts who study sleep-trackers say that while there are some benefits to knowing this data, it can be presented in misleading ways, and they caution that sleep trackers aren’t a cure for insomnia or other sleep-disorders. Here’s what to know about the trackers’ capabilities and limitations.

How They Work Many wearables collect data using something called photoplethysmography, or PPG.

“There’s this little light on the back of the device that shines into the blood-vessels, and it uses the amount of light that is reflected-back to estimate things like heart-rate and heart-rate-variability,” said Dr. Cathy Goldstein, a Clinical Professor of Neurology at the University of Michigan who studies consumer sleep-monitors.

She added that most wearables, and also many nearables, contain accelerometers that can measure motion.

Sleep-trackers that are placed under a mattress (or, in the case of some “smart-beds,” inside the mattress) often rely on ballistocardiography, a technology Dr. Goldstein said is capable of detecting subtle movements caused by the heart’s pumping-action — movements that correlate with sleep and its stages.

Researchers have found that the latest sleep-trackers are generally adept at detecting the basics: when a person is asleep or awake. However, Dr. Goldstein said that trackers can be less-accurate when collecting data from people with obesity or heart rhythm disorders such as atrial fibrillation, as well as from those who have darker skin tones, because skin pigment can interfere with the way light is reflected back to the device.

But even if the data collection is perfect, experts say that many of these technologies overreach when they attempt to translate findings into consumer-friendly takeaways.

“They’re presenting information with a granularity that they’re not yet capable of,” Dr. Goldstein said.

For example, while many trackers offer data on a user’s sleep stages, such as REM sleep and deep sleep, these stages are defined by shifting patterns of brain activity — something most devices can’t directly measure.

“Inferring sleep and the sleep stages from peripheral phenomena like pulse rate or respiration has some inherent limitations, especially if the person is not healthy,” said Mathias Baumert, an associate professor of biomedical engineering at The University of Adelaide in Australia, who specializes in health technology.

It’s also unclear how people could benefit from this information. “We don’t diagnose sleep disorders based on REM sleep or deep sleep,” said Kelly Baron, a clinical psychologist and director of the Behavioral Sleep Medicine program at the University of Utah. She said that even people who sleep well have differing patterns of REM or deep sleep as a result of age, sex, medication use and other variables.

“I have patients who come in saying they’re worried because their device is telling them they’re not getting enough deep sleep, but I couldn’t even tell you how much deep sleep is optimal,” she said.

The experts are especially critical of sleep trackers’ attempts to aggregate a person’s nightly data into an overall sleep grade or score. In a 2022 paper, Dr. Baumert and his coauthors pointed out that the algorithms companies use to determine these scores are often proprietary and not scientifically vetted.

“A simple metric is attractive from a consumer perspective,” Dr. Baumert said. “But it is difficult to understand what is measured and what those scores mean in terms of health outcomes and disease.”

Dr. Goldstein put it more bluntly: “Ugh, these sleep scores or readiness scores are the worst. I tell my patients to ignore those.”

How to Use Them One of the advantages of sleep trackers is their ability to capture and record long-term data in a person’s natural sleeping environment. “No matter how sensitive the equipment, a night spent in a sleep lab is not representative of a hundred nights of sleep at home,” Dr. Baron said.

The potential to spot meaningful long-term patterns in one’s sleep — the way Mr. McCall, the personal trainer, noticed how alcohol and late-night meals messed with his slumber — is “extremely exciting, both for sleep scientists and for people who have these devices,” Dr. Baron added.

Dr. Goldstein said the data from these devices could also reinforce the benefits of sleep hygiene. For example, a user could see how going to bed and getting up at the same time each day positively affects their metrics.

On the other hand, people who are already anxious about their sleep may want to think twice before using a tracker. Research on real-world users has found that these devices can stress people out or increase their focus on sleep, which can be counterproductive.

“If you’re not sleeping well, having this device that tells you how poorly you’re sleeping might make things worse,” Dr. Goldstein said.

Finally, it’s important to remember that there’s a lot about sleep that remains a mystery. “There is still so much we need to learn about the role of sleep, and how patterns of sleep and sleep disruption affect health,” Dr. Baron said.

“I think the current devices can be fun for people and provide some interesting information,” she added. “But sleep can’t be boiled down into a set of numbers or scores.”

Q: What is the best device for accurately measuring sleep and its stages at home?

[ChatGPT4o] Accurately measuring sleep-stages at home has become increasingly feasible with advancements in consumer-grade sleep-tracking devices. While these devices may not match the precision of clinical-polysomnography, several have demonstrated commendable accuracy in tracking sleep-patterns and stages.

Top Devices for Home Sleep-Stage Monitoring

Oura Ring 4 A sleek, lightweight ring that monitors sleep-stages, heart-rate, and body-temperature. It provides detailed sleep-analysis and personalized-insights.

The Verge

WHOOP 4.0 A wrist-worn device that tracks sleep-stages, strain, and recovery-metrics. It's favored by athletes for its comprehensive data and coaching feedback.

Health.com

Muse-S Headband A comfortable headband that monitors brain-activity, heart-rate, and breathing-patterns during sleep. It offers real-time feedback and guided sleep meditations.

Sleep Foundation

Withings Sleep Analyzer A non-wearable device placed under the mattress that tracks sleep cycles, heart rate, and snoring. It provides a comprehensive sleep score and integrates with health apps.

Sleep Foundation

Garmin Venu 3 A smartwatch offering advanced sleep-monitoring, including sleep-stages, Pulse Ox, and respiration tracking. It also provides health and fitness metrics.

The Verge

Considerations When Choosing a Sleep-tracking Device:

• Accuracy: While consumer devices have improved, they may not fully replicate clinical sleep studies. However, many users find them sufficiently accurate for personal sleep pattern monitoring.
• Comfort: Select a device that suits your comfort preferences, especially if you are sensitive to wearing accessories during sleep.
• Data Insights: Consider the depth and usability of the sleep data provided, including how it integrates with other health metrics.
• Battery-Life: Ensure the device's battery life aligns with your usage patterns to avoid frequent charging interruptions.
• Subscription-Fees: Be aware of any additional costs for accessing premium features or detailed analytics.

While no consumer device can entirely replace the accuracy of a clinical-sleep-study, the aforementioned devices offer valuable insights into sleep patterns and stages. They serve as practical tools for individuals seeking to monitor and improve their sleep quality at home.

The Best Sleep-trackers You Can Buy: Expert Tested

We tested the top sleep-trackers that collect data throughout the night to help you further understand your sleep.

by Nina Raemont, Associate Editor - Reviewed by Kayla Solino

ZDNET's reviewers spend weeks to months testing each item on this list. We aim to give you a view into how each item could actually fit into your life and workflow.

Best sleep-tracker overall

Oura Ring Gen 4

View at Amazon

Apple Watch Series 10 | Best sleep-tracker for Apple obsessives

Best sleep-tracker for Apple obsessives

Apple Watch Series 10

View at Amazon

Google Pixel Watch 2 | Best sleep-tracker for Android

Best sleep-tracker for Android

Google Pixel Watch 2

View at Walmart

Garmin Venu 3S | Best sleep-tracker for athletes

Best sleep-tracker for athletes

Garmin Venu 3S

View at Amazon

Whoop 4.0 | Best screenless sleep-tracker

Best screenless sleep-tracker

Whoop 4.0

View at Amazon

A sleep-tracker might not get you to sleep quicker, make your slumber longer, or keep you asleep for the entirety of the night, but it will help you understand your sleep better. That's what a good sleep-tracker should do, anyway. Once you can see how well (or poorly) you slept you can make goals to improve it, change aspects of your routine or diet, and maybe even get enough exercise throughout the day to be tired enough to fall asleep before midnight -- talk about a dream. 

Experts recommend that adults get at least 7 hours of sleep every night (though everyone is different). Even with 7 hours of sleep, there's no guarantee that we wake up well-rested. To determine how you've actually slept throughout the night, sleep-trackers and wearables come in handy. While not perfect, sleep-trackers can measure heart-rate, skin-temperature, and time-spent in each sleep-phase. 

There are smartwatches, smart rings, and earbuds that track your sleep to provide a greater picture of your sleep health. Your sleep-data will differ depending on which product you use, and some products perform their sleep-tracking duties better than others. The ZDNET team has carefully tested sleep-trackers to find the best one based on features, affordability, and availability. 

Also: I tested the Oura Ring 4 for weeks. Here's why it's the best health tracker right now

What's the best sleep-tracker right now?

We found that the Oura Ring 4 is the best sleep-tracker on the market for its comfortable, lightweight-fit, impressive battery-life, discreet build that makes it perfect for sleeping. Plus, it offers a plethora of sleep data on the app. But the Oura Ring isn't the only sleep-tracker to consider. Here, we explore the most-popular options, as well as some you may not have heard about, to help you choose the best sleep-tracker for a better night's sleep.

The Best Sleep-trackers of 2024

View now at Amazon

Pros

• To-the-tee sleep-tracking
• Data illustration that informs your sleep and activity routines
• Comfortable build for all-day-wear

Cons

• Paid membership required to unlock all data the ring collects
• Battery-life isn't as long as advertised

Oura Ring Gen 4

Best sleep-tracker overall

The Oura-Ring-4 is a competitive sleep-tracker, with insights into everything from sleep-efficiency, restfulness, readiness, latency, and more. I've been using the Oura Ring for a while now, and it's not only given me deeper insight into my sleep but also supplements those data points with explanations behind the numbers. Oura claims to accurately track sleep-cycles, total-sleep-time, resting-heart-rate, heart-rate-variability, and movement. 

These features have been enhanced through the rollout of the fourth-generation Oura Ring, which hit the market in October. The Oura-Ring-4 has smart-sensors, allowing more-accurate and continuous data capture. It's also more comfortable than the Oura-Ring-3 and has flattened-sensors that make it easier to wear over time. 

Every morning, I get a sleep-score for last night's sleep and view my heart-rate, my sleep-efficiency, and my suggested bedtime range for the day. The sleep-score helps me understand my energy levels for the day and my activity bandwidth, like whether I should exercise or rest. But it's not just deep-sleep that it can track and point-out; the Oura Ring can detect the shortest of naps and factor those into your sleep score for the day. Plus, the sleep-tab on the app offers information on sleep-efficiency, restfulness, REM, deep-sleep, latency, and timing, so you can dig deep and go beyond viewing only a sleep-score and your time spent in bed. 

As if there aren't enough sleep-friendly features through the Oura Ring, Oura also unveiled a body clock feature that teaches you more about your circadian rhythms and body clock. The feature measures my sleep chronotype, calculated by my body temperature, sleep-wake cycle, and physical activity, and through this data gathered that I'm most productive during the late morning. 

The battery-life is masterful. It charges quickly and stays charged for around 5 to 6 days. Oura says that the new generation's battery life lasts 8 days, but I haven't found that to be the case in my testing. The one downside to the ring is the extra $72 you spend on the Oura subscription a year, on top of the $350 you just spent on the ring itself. 

Redditors also agree that the Oura Ring, as a sleep-tracker, is worth it. "If you struggle with sleep quality, overtraining, or anything else that affects recovery, I'd say it's worth it," one user wrote in a thread. However, one Best Buy reviewer noted that the ring doesn't offer half-sizes, which could be an issue if your fingers swell.

Oura Ring Gen 4 specs: Up to 8 days battery life | Sleep-cycles | Bluetooth | Up to 100m water-resistant | Blood-oxygen, skin-temperature sensor, heart rate monitoring | iOS and Android compatible

View now at AmazonView now at Best Buy

Pros

• Sleep cycle tracking
• No subscription to track sleep
• Sleep apnea detection

Cons

• Must be paired with iPhone
• Shorter battery life than other wearables
• No sleep scores

Apple Watch Series 10

Best sleep-tracker for Apple obsessives

I've tried to sleep with an Apple Watch around my wrist, and while the tech does a fairly good job at tracking my sleep, even the lightest smartwatch pales in comparison to a smart ring for sleep-tracking functionalities. That being said, I know there's a lot of people who live and die by their Apple Watch, and if that's you, the Apple Watch Series 10 is the best you can get for sleep-tracking. 

The Apple Watch Series-10 offers many improvements from its Series 9 predecessor. Not only does the watch come with a bigger display, but it's also brighter and thinner. One key new feature of the Series-10 is Apple's FDA-cleared sleep-apnea-detection. People can wear the Series-10 to bed and get information about breathing disturbances throughout the night, then take that data to the doctor for a diagnosis. This data can be viewed on a one-month, six-month, or one-year timeline, available in the Health app. 

Review: Apple Watch Series 10

With the Sleep app on Apple Watch, you can create goals for how long you'd like to sleep each night and adjust them for weekdays and weekends, use the Sleep Focus mode, which limits distractions before bed, and track your sleep through the motion-detectors on the watch. It estimates how long you spend in each sleep-phase -- REM, core, and deep – per-night. Note that Sleep Focus needs to be on to begin tracking your bedtime activity. 

My least favorite thing about the Apple Watch is that you don't get a sleep-score or summary that helps evaluate your night's sleep. A number rating is far easier to understand than a heart-rate-graph, so hopefully, one day, Apple will catch-up to the competition with a reliable sleep and readiness score-system. 

Apple Watch Series 10 specs: Up to 18-hour battery life | Sleep cycles and sleep apnea-tracking | Wi-Fi, Bluetooth, GPS, and/or cellular options | 42mm and 46mm models | IP6X dust and water resistant | Blood oxygen, body temperature sensor, EKG, irregular heart rate alerts | iOS only

View now at WalmartView now at Crutchfield

Pros

• Comfortable and light
• Fitbit integration
• Confirmed accurate heart rate and GPS tracking

Cons

• Limited exercise details
• Only one size
• Smaller battery life

Google Pixel Watch 2

Best sleep-tracker for Android

Android owners looking for a sleep-tracker on a smartwatch should look no further than the Google Pixel Watch 2. Google improved its heart rate sensors in its newest iteration of the Pixel Watch, which ZDNET contributing writer and resident smart watch expert Matt Miller confirmed when he tested out the watch himself. 

On top of that, the Google Pixel Watch 2 further integrates Fitbit into the wearable with improved stress tracking, a daily readiness score, ECG, sleep scores, and six months of Fitbit Premium for free. Some of Fitbit's sleep-tracking features include nightly trends and trends over time, a silent alarm that vibrates to wake you up during your most optimal sleep stage, bedtime reminders, sleep goals, and more. One Best Buy reviewer said that the watch's silent alarm makes it easy to wake up without waking her husband, and a handful of reviewers pointed out the watch's accurate sleep-tracking.  

If you're wearing a piece of tech to bed, it needs to be lightweight and compact. The Pixel Watch 2 is much lighter than its predecessor, ideal for comfortable sleep. "Google's continued focus on sleep and stress tracking means having a lighter, smaller watch makes more sense," Miller writes in his review. 

More: Google Pixel Watch 2 Review 

While the 24-hour battery life isn't nearly as competitive as other smart watches, the faster charging rate takes your 0% battery to 50% in less than 30 minutes. Miller says it's still one of the most accurate Fitbit-powered wearables for heart rate and GPS tracking, so what you lack in battery life you gain in sublime accuracy. 

Google Pixel Watch 2 specs: 24 hours battery life | Sleep stages | 41mm | Up to 164 feet of water resistance | Accelerometer, Altimeter, Compass, Multi-path Optical Heart Rate, Blood Oxygen Sensor, Skin Temperature, Barometer, cEDA, Gyroscope, Light Sensor | Android compatible 

Show less

View now at AmazonView now at Garmin

Pros

• Sleep coach
• Strong battery life
• Great for athletes and fitness junkies

Cons

• Sleep-tracker significantly drains battery, users report

Garmin Venu 3S

Best sleep-tracker for athletes

If you're an athlete or fitness junkie shopping around for a new smart watch, you've probably heard about Garmin's smart watches and their suite of health tracking features, like the body battery, workout benefit and recovery time, and perceived exertion data. Most runners I know use one of Garmin's watches for their half marathon and marathon training, and our resident smart watch expert, Matt Miller, loves them for their rugged build. So, if you're all-in on your workout regimen, a Garmin watch that's already giving you accurate sport insights that pairs with special sleep capabilities might be your sleep-tracker of choice. 

Out of all the Garmin watches, the Venu-3S might be the brand's best for sleep-tracking due to its lighter and smaller build and its sleep coach feature, which provides you with a sleep score, tracks your sleep stages and naps, and monitors your heart rate variability and skin temperature during rest. All this sleep-tracking data impacts your body battery, which Garmin delivers to you in a score out of 100. 

The sleep coach combines your age, daily and longer-term activity levels, recent sleep history, naps taken, and heart rate variability (HRV) to offer you a recommended nightly sleep need. 

The Venu 3S is of a smaller build than other Garmin watches, which works wonderfully for sleep. The smooth silicone band of the watch also resembles that of the Apple Sport band, according to Miller. Miller's daughter, an avid hiker and napper, tried it out and enjoyed the nap tracking feature on the watch. 

The only downside of this sleep-tracker is how it significantly drains the watch's battery, some Amazon reviewers have complained. But even if the battery is drained, its long battery life still out competes many sleep-trackers on this list, and is great for anybody who wants to improve their fitness quality through sleep-tracking. 

Garmin Venu-3S specs: Up to 10 days battery life | Sleep score and insights | Bluetooth | 41mm or 45mm | 5 ATM water resistant | GPS, heart rate, pulse oximeter blood oxygen monitor, barometric altimeter, accelerometer | iOS and Android compatible

View now at AmazonView now at Best Buy

Pros

• Great customer service, users report
• Detailed sleep-tracking for athletes
• Personalized sleep coach

Cons

• Annual membership subscription

Whoop 4.0

Best screenless sleep-tracker

If you want a lightweight, screen-less band that effortlessly tracks all your activity and sleep data, your best bet could be the Whoop 4.0. Countless professional athletes have used the smart band, given its focus on measuring strain and recovery. But where it really shines is in its sleep-tracking. 

Whoop measures your four sleep stages (slow wave, awake, REM, and light) and suggests optimal sleep times based on the sleep data it collects. You can also set alarms that go off at a specific time, once you hit your sleep goal, or once you've hit a recovery period. One of the best sleep features Matt Miller, who reviewed the Whoop 4.0 pointed out, was the sleep coach that not only informs you when to go to bed but tells you how much sleep you will reach if you go to bed past that time. Plus, the Whoop calculates how much sleep you'll need based on the energy you exert on physical or mental tasks. 

You get all your health data through the Whoop smartphone app, given the fact that this band is screen-less. Launching the app displays data on strain, recovery, and sleep. To go deeper into your sleep data, you can get a by-the-minute plot of your heart rate and other sleep metrics during the night, along with some words that describe the data presented for added clarity. 

You can analyze your daily, weekly, and monthly sleep patterns through the detailed sleep trends Whoop offers and calculate your true sleep need, the amount of time you should be sleeping a night. "At first, I thought the Whoop was just an advanced sleep-tracking device that offered more than my GPS sports watches, but it's more than that. It does track the details of your sleep but is also measuring other key metrics of your body throughout the day," Matt Miller writes in his review of the Whoop 4.0. 

Customers agree and say the Whoop is worth the money. Many Amazon reviewers said they initially bought the Whoop 4.0 to track their sleep but found themselves improving other areas of their life thanks to the detailed data and health metrics that Whoop offers. 

Review: Whoop 4.0 

Whoop 4.0 specs: Up to 5 days battery life | Sleep cycles | Bluetooth | 1.5 x 1.5 x 0.45 inches | IP68 water resistant | Blood oxygen, skin temperature sensor, heart rate | iOS and Android compatible

What is the best sleep-tracker?

The best sleep-tracker overall is the Oura-Ring-4 thanks to its suite of features that help you understand your health data behind your sleep-score. It's more-comfortable to wear compared to a smartwatch and offers the most in-depth summaries and recommendations of your night's slumber. 

Best sleep-tracker	Cost	Battery life	iOS/Android	Subscription
Oura Ring 4	$350	8 days	iOS and Android	Yes
Apple Watch Series 10	$399	18 hours	iOS only	No
Google Pixel Watch 2	$350	24 hours	Android only	No
Garmin Venu 3S	$450	10 days	iOS and Android	No
Whoop 4.0	$240	5 days	iOS and Android	Yes

Show less

Which is the right sleep-tracker for you?

That depends on what you need. Do you want something with a screen or no screen? Are you an athlete looking to combine your sleep data with your training data or someone who's casually interested in their sleep score night by night? We've compiled a list of picks that will suit all sorts of needs and use cases. 

Choose this best sleep-tracker...	If you want...
Oura Ring	The best sleep-tracker overall, with a battery life that lasts nearly a week, comprehensive sleep data, and a form-factor you won't ever notice.
Apple Watch Series 10	The best sleep-tracker for Apple users with easy Apple integration, sleep cycle information and sleep apnea detection, and a larger, thinner, and brighter display.
Google Pixel Watch 2	The best sleep-tracker for Android users, with Fitbit integrated sleep tech, a daily readiness score, and a lightweight build that makes the Pixel Watch 2 easy to sleep with.
Garmin Venu 3S	The best sleep-tracker for athletes or fitness junkies. It's a smart watch, so you can take calls on it, check your mileage as you run, and get in depth data on your sleep. The Venu 3S has the highest battery life on this list, and can even detect the shortest of naps.
Whoop 4.0	The best sleep-tracker for athletes who don't want a screen on their wrist. Get insights and recommendations through Whoop's personalized sleep coach and learn how you can recover from your workouts with the right amount of sleep that Whoop calculates for you.

Factors to Consider When Choosing a Sleep-tracker

Poor sleep quality can lead to an increased risk of dementia, obesity, cardiovascular disease, depression, and even certain types of cancer. Sleep has never been more important. 

To find the best sleep-trackers, ZDNET relied on hands-on testing and thorough research into product quality as well as sleep studies on these devices. Additionally, we considered the following factors:

• Sleep-tracking: Sensors in your sleep-tracker are capable of collecting a variety of data, such as your heart rate, heart rate variability, time asleep, sleep stages, and even how much you snore. All of this data is collected and analyzed, and the best sleep-trackers present this information to you with clarity. 
• Comfort: If you're wearing a device to bed, it should be easy to sleep with. We've tried to consider smart watches, bands, and rings that are lightweight, compact, and unobtrusive. 
• Additional features: Not all sleep-trackers stop at just sleep-tracking. Others may also provide additional features, like a smart alarm or health and fitness tracking. These insights can help you live a better life during the day so you sleep better at night. 
• Subscription: In order to access some features, you may need to subscribe to a plan. Some subscriptions may require a one-time fee, while others may require that you pay an ongoing monthly fee for the service. Be sure to check out the terms of the subscription before you buy. Although they are generally pretty affordable, this is an additional cost that you will need to factor into the total price. 
• Price: Sleep-trackers can be reasonably priced, but they start to get pretty pricey once you start to add extra features. I found that the cost of the best sleep-trackers range in price from $80 to $400, depending on the model you choose.  
• User sentiment: We read the customer reviews on each product to see what other people were liking and disliking about each sleep-tracker. You should too.

How did we choose these sleep-trackers?

We know how much of an investment any type of personal tech can be. That's why when we pick products to include on a list like this, we give these products careful consideration. We do this by rigorously testing them for days and weeks on end, reading customer reviews, and evaluating whether they live up to their value and cost. 

The ZDNET team is constantly testing and ranking products, swapping one old model on the list for a new model, and getting rid of products that are no longer competitive with the others on the list. We take this testing seriously because whether a product costs $30 or $300, we want you to get your money's worth. 

What is a sleep-cycle?

A sleep-cycle is comprised of three different stages: light-sleep, deep-sleep, and rapid-eye-movement, or REM. On a typical night, when you first fall asleep, you'll enter either light- or deep-sleep, likely one after the other, and enter a REM-phase. 

This cycle will last about 1.5-hours, then repeat itself in intervals closer to 2-hours, maybe with a few awake minutes in between, but always ending with REM-sleep. As the night progresses, you'll likely experience less time in deep-sleep, while the REM stages will become longer. 

A regular night of sleep should be about 50% light-sleep, 20% REM-sleep, 20% deep-sleep, and 10% awake. 

How can I improve my sleep? The National Sleep Foundation recommends improving your sleep by practicing certain habits throughout the day and before bed. Following a consistent bed routine, that is, waking-up and retiring at around the same time, is one way to signal to your body to ease into sleep. Dimming the lights a few hours before bed, limiting screen-time, and creating a cozy, dark, and cold bedroom environment are all ways to improve your sleep-schedule. Regular-exercise and limited alcohol, caffeine, and nicotine consumption can also help keep you asleep. 

How much REM do I need? Nearly a quarter of your sleep should be REM-sleep, according to health experts. So, if you're sleeping 8-hours a night, the recommended average, 2-hours of REM each night should be ideal. 

Are sleep-trackers really accurate? If you need your sleep analyzed, doctors will usually order a polysomnography, which analyzes a person's oxygen-level, heart-rate, breathing, eye-movement, and leg-movement while simultaneously running an electroencephalogram (EEG). 

An EEG shows your brainwaves through different stages and cycles of sleep. A wearable smartwatch or fitness-tracker on your wrist, however, is far from your brain.

The most popular wearables by Fitbit, Apple, Garmin, and Samsung collect a combination of the following measurements: heart-rate and heart-rate-variability (HRV), movement, breaths-per-minute, and skin-temperature. Devices that track sleep-cycles use an algorithm to hypothesize your sleep-cycles based on the data.

A 2017 study compared a polysomnography with the performance of a Fitbit-Charge-2 in 35 adults. Researchers found the Fitbit device detected sleep onset with 96% accuracy but overshot time spent asleep by 9 minutes on average. In the study, the Fitbit detected light sleep with 81% accuracy, deep-sleep with only 49% accuracy, and REM-sleep with 74% accuracy. 

A newer study published in April of 2022 compared the performance of Fitbit-Alta-HR to results of an EEG conducted simultaneously on 40 college athletes while sleeping. This study found FItBit data to satisfactorily track sleep-onset, time-spent-asleep, and sleep-cycles. Fitbit can be a useful tool for athletes' sleep management.

While not 100% accurate, the Fitbit data had a strong correlation with polysomnography-data. 

What types of sleep-trackers are there?

There are several types of sleep-trackers that you can choose from. Smartwatches, rings, and under mattress nearables are the three most popular options. 

• Smartwatches: These can be either smartwatches or wristbands, depending on the amount of technology that you need. Smartwatches often incorporate smart assistants like Alexa or Google Assistant while offering a wealth of features like GPS tracking and fitness tracking in addition to your regular sleep-tracking.
• Rings: This is another wearable sleep-tracker but far more basic. It uses sensors along the inside of the ring to collect data and track your sleep. 
• Under-mattress or nearables: There are some models, like the Withings Sleep-tracker Pad, that simply slide under your mattress for an unobtrusive solution to sleep-trackers.

What are the best sleep-trackers that don't require a subscription?

The Apple Watch Series 9, Google's Pixel Watch 2, and the Garmin Venu 3S are all impressive smart watches that offer great sleep-tracking capabilities -- subscription free. If you're a casual sleep-tracker, I'd recommend the Apple Watch for people within the Apple ecosystem and the Pixel Watch 2 for those in the Android ecosystem. If you're an athlete or someone who's training for a marathon I'd suggest the Garmin Venu 3S. 

2024's Top Rated Smart Health Rings Tested: One Device Redefines Wearable Health Tracking

December 24, 2024

Are you tired of bulky fitness trackers that interfere with your daily activities?

Do you want comprehensive health monitoring in a sleek, minimalist design?

Are you looking for a discreet way to track your fitness and wellness 24/7?

We've meticulously tested the top Smart Health Rings from Oura, Circular, BodiMetrics, and other innovative brands available at major retailers like Best Buy, Amazon, and specialty fitness stores.

Our testing revealed remarkable advancements in wearable technology, with one device that truly stood out from the pack. This fitness-ring exceeded our expectations, redefining what we thought possible in compact health-tracking.

The highest-rated ring comes from a company known for pioneering smart wearable solutions. Their cutting-edge approach has produced a fitness-ring that excels in accuracy, battery-life, comfort, and comprehensive health-insights. 

#1 TITAN Longevity Ring

• Tracks 20+ health metrics with medical-grade accuracy & research-grade tech
• Lightweight, high-strength & hypoallergenic titanium alloy crafted for seniors
• 24/7 monitors sleep-quality, heart-rate, blood-oxygen & stress-levels
• Make you sleep smarter with advanced tracking
• Offers the highest accuracy in measurements with advanced fingertip sensors
• Sleek & modern aesthetic: a stylish accessory for any occasion
• 5atm waterproof: perfect for swimming, showers, or workouts
• Ai-powered insights via intuitive app: make informed health decisions
• Impressive battery life (up to 7-days): stay powered without frequent charging
• Durable & scratch-resistant finish: built to withstand daily wear and tear

• High demand – often sells out quickly

overall score - 9.9 - 4,200+ users gave this brand 5-Stars

Visit Site

 

50% OFF - Christmas Sale Starts December 24, 2024 - $59.99 $133.31 - 1-3 Days Express Shipping

2

• Monitors 10+ Vital Signs with Precision Accuracy
• Tracks Sleep Quality, Heart Rate, and Activity Levels
• Durable Build: Waterproof & Scratch-Resistant
• Comfort-Fit Design: Wear It All Day Without Discomfort
• Chic Aesthetic: Blends Seamlessly with Any Outfit
• Long-Lasting Battery: Enjoy 7 Days of Continuous Tracking
• Smart Insights Delivered Through an Intuitive App
• Limited Stock

overall score - 9.8 - 3,200+ users gave this brand 4.5 Stars

Visit Site

$53.99 - 2-5 Days Standard Shipping

3

• Customize fit with sizing kit
• Enjoy first month free membership
• Track 30+ health metrics
• Experience up to 8 days battery
• Wear comfortably 24/7
• Sync effortlessly with apps
• High price point
• Not the most attractive appearance

overall score

9.3

1,900+ users gave this brand 4 Stars

Visit Amazon

$346

2-5 Days Standard Shipping

4

• Achieve perfect fit with sizing kit
• Enhance sleep with actionable insights
• Track all-day activity effortlessly
• Recover better with dynamic scoring
• Monitor health with sleek design
• Align rhythms for optimal health
• Not the best value for money
• Lack of online reviews

overall score - 9.0 - 840+ users gave this brand 4 Stars

Visit Amazon

$296 1-3 Days Express Shipping

5

• Ensure perfect fit with sizing kit
• Track health metrics effortlessly
• Experience precision biometrics monitoring
• Integrate seamlessly with apps
• Enjoy durable, lightweight design
• Monitor health discreetly in style
• Customer service issues
• Not the most user friendly option

overall score - 8.9 - 720+ users gave this brand 3.5 Stars

Visit Amazon

$249 - 1-3 Days Express Shipping

27 Expert Reviewers

45,000+ Reviews Checked

120+ Smart Health Rings Tested

CGT Recommends: TITAN Longevity Ring

Pros

• Tracks 20+ health metrics with medical-grade accuracy & research-grade tech
• Lightweight, high-strength & hypoallergenic titanium alloy crafted for seniors
• 24/7 monitors sleep quality, heart rate, blood oxygen & stress levels
• Make you sleep smarter with advanced tracking
• Offers the highest accuracy in measurements with advanced fingertip sensors
• Sleek & modern aesthetic: a stylish accessory for any occasion
• 5ATM waterproof: perfect for swimming, showers, or workouts
• AI-powered insights via intuitive app: make informed health-decisions
• Impressive battery-life (up to 7 days): stay powered without frequent charging
• Durable & scratch-resistant finish: built to withstand daily wear and tear

Cons

• High Demand – Often Sells Out Quickly

Visit Site

 

50% Off - Christmas Sale Starts: December 24, 2024

After an intense testing period, TITAN Longevity Ring exceeded all of our expectations, making it the top choice for holistic health tracking in 2024.

This sleek, lightweight ring packs powerful fitness-tracking features into a stylish design that's perfect for everyday wear. It tracks everything from heart rate changes to blood oxygen saturation and real-time workout data, making it easy to monitor your health throughout the day. Whether you're running, walking, or simply going about your daily routine, this ring ensures you're always in the know.

What really stood out to us was the ring's sleep-tracking abilities. It accurately recorded sleep-stages, including REM, and provided insightful feedback to help improve sleep-quality within just a week of use. Its waterproof design also means you don't need to worry about taking it off in the shower or during intense workouts.

The TITAN Longevity Ring syncs effortlessly with its app, where you can view detailed daily, weekly, and monthly reports. You can also easily share this data with healthcare providers, adding another layer of health insight.

Despite its premium features, the TITAN Longevity Ring is surprisingly affordable. With its current 50% discount, it's an deal for anyone looking to improve their fitness and overall well-being.

How Your Sleep Patterns Change Can Tell You About Your Health

Researchers Identified 5 Sleep Types; The Way People Move Between These Types Yields Insights About Both Chronic and Acute Conditions

 

June 20, 2024 - University of California - San Diego - The study appeared in the journal npj Digital Medicine 

Summary: Your sleep-tracker might give you information about more than just your sleep -- specifically, it might give you information about chronic conditions such as diabetes and sleep-apnea, and illnesses such as COVID-19. This is one of the findings of a study that analyzed data from 5-million-nights-of-sleep across roughly 33,000 people.    

 

This is one of the findings of a study that analyzed data from 5-million nights-of-sleep across roughly 33,000 people. Based on the new analyses, the researchers identified 5 main types of sleep, which they called sleep-phenotypes, and which can be further divided into 13-subtypes.

 

The researchers also found that how and how often a person switches between sleep phenotypes could offer 2- to 10-times more-information relevant to detecting health conditions compared with just relying on a person's average sleep phenotype alone.

Using data collected from Oura-Ring a smart-ring that tracks sleep, skin-temperature and other information-the researchers looked at individual people over a series of months, noting whether they had chronic health conditions such as diabetes and sleep-apnea, or illnesses such as COVID-19 and the flu.

The research team found that people would often move between sleep-phenotypes over time, reflecting a change in an individual's health conditions, and creating what resembles a person's travel-log through the data-driven sleep-landscape the researchers created.

"We found that little changes in sleep-quality helped us identify health-risks. Those little changes wouldn't show up on an average night, or on a questionnaire, so it really shows how wearables help us detect risks that would otherwise be missed," said Benjamin Smarr, one of the study's senior authors and a faculty member in the Jacobs School of Engineering and Halicioglu Data Science Institute at the University of California San Diego.

In addition, the researchers highlighted that tracking changes in sleep over the long-term at the population-scale could unlock new insights that are relevant for public-health, such as whether some changes in patterns through these sleep-landscapes can provide early warning for chronic-illness or vulnerability-to-infection, [or even possibly future dementia.]

The research team's work is based on new analyses from the TemPredict dataset from University of California, San Francisco, which was created using data collected from people wearing the commercially available Oura Ring during the 2020 COVID-19 pandemic.

The analyses were led by Smarr, who is also faculty in the University of California San Diego Shu Chien -- Gene Lay Department of Bioengineering, and Professor Edward Wang in the University of California San Diego Department of Electrical and Computer Engineering, collaborating with the study lead at University of California, San Francisco, Professor Ashley E. Mason, a practicing sleep-clinician. The lead author was Varun Viswanath, a graduate student in the Department of Electrical and Computer Engineering at the University of California San Diego Jacobs School of Engineering.

The 5 Sleep-Types These are the 5-sleep-phenotypes researchers singled-out based on data from 5-million nights of sleep across roughly 33,000 people. While many factors went into the study, the researchers also identified some trends that help to intuitively separate the 5 sleep-phenotypes.

• Phenotype-1: What we think of as "normal" sleep. In this phenotype, people get about 8-hours of uninterrupted-sleep for at least 6-days in-a-row. This is the type of sleep recommended by the National Institutes of Health and was the most-common sleep type researchers found.
• Phenotype-2: People sleep continuously about half the nights, but they only sleep for short periods of time in bouts of less than three hours the other-half.
• Phenotype-3: People sleep mostly continuously, but they experience interrupted sleep around one night each week. The interrupted night is characterized by one period of relatively long sleep of about 5-hours, and one period of short-sleep of less-than-3-hours.
• Phenotype-4: People again sleep mostly continuously. But they experience rare nights, in which long bouts of sleep are separated by a mid-sleep-waking.
• Phenotype-5: People only sleep for very-short-periods-of-time every-night. This phenotype was the rarest the researchers found and represents extremely-disrupted-sleep. [Unfortunately, I am one of those with a fragmented, short-sleep-intervals, pattern. Need to have a diagnosis with a polysomnography (sleep-study): Conducted in a sleep-lab or at home (home-sleep-apnea-testing) that monitors breathing-patterns, oxygen-levels, heart-rate, and sleep-stages.

Tracking Changes in Sleep-Type To measure how sleep-phenotypes changed over time, Viswanath constructed a spatial-model of all 5-million nights, in which the phenotypes were represented as different islands, composed of mostly-similar weeks-of-sleep. Different patterns emerged over time that allowed the researchers to model each individual's routes between islands.

From there, what helped to distinguish people with chronic-conditions, such as diabetes and sleep-apnea, was not their average phenotype. Instead, it was how frequently they switched between islands in this sleep-landscape. In this way, even if someone switched phenotypes only rarely, the fact that they did switch could still provide useful information about their health.

The data showed that it is rare for most people to go multiple months without a few nights of disrupted sleep. "We found that the little differences in how sleep-disruptions occur can tell us a lot. Even if these instances are rare, their frequency is also telling. So, it's not just whether you sleep well or not -- it's the patterns-of-sleep-over-time where the key info hides," said Wang, a Coauthor and Electrical and Computer Engineering Faculty Member at UC San Diego.

Conversely, people did not tend to remain in patterns defined by broken-up-sleep. But how often they visited specific disrupted-sleep-patterns says a lot about how well they're doing.

"If you imagine there's a landscape of sleep-types, then it's less about where you tend to live on that landscape, and more about how often you leave that area," said Viswanath, the paper's corresponding author.

Previous Research In this new paper published June 20, the research team modified the technique used in previous research that had been the largest similar investigation of sleep to date, which had drawn approximately 103,000 nights of data from the UK biobank. That previous study looked at sleep timing and awakenings and many related features, and then constructed a "landscape" of where nights fell in relation to each other. But prior researchers did not do two key things: they could not look across-time, as they had only 2-to-3-nights-per-person; and they could not tie the resulting patterns-of-sleep to health-outcomes.

Other large-scale sleep-analyses looked at high-level-differences in simple-sleep-characteristics, like the total-time people spent-asleep.

In contrast, this new work is the first to show that researchers can quantify the changing dynamics of people's sleep over time and use this quantification to give people better insights into their sleep-health. The research also suggests that these changes in sleep may indicate a higher risk for a wide range of conditions.

Story Source: Materials provided by University of California - San Diego. Original written by Daniel Kane. 

Journal Reference: Varun K. Viswanath, Wendy Hartogenesis, Stephan Dilchert, Leena Pandya, Frederick M. Hecht, Ashley E. Mason, Edward J. Wang, Benjamin L. Smarr. Five Million Nights: Temporal Dynamics in Human Sleep Phenotypes. npj Digital Medicine, 2024; 7 (1) DOI: 10.1038/s41746-024-01125-5

University of California - San Diego. "How your sleep patterns change can tell you about your health." ScienceDaily. ScienceDaily, 20 June 2024. <www.sciencedaily.com/releases/2024/06/240620152316.htm>.

Electroencephalography (EEG) Signal Analysis

 

Advancements in Electroencephalography (EEG) analysis have significantly enhanced our understanding of brain-function, particularly in diagnosing and treating neurological-disorders. Contemporary techniques integrate sophisticated computational-methods and machine-learning-algorithms to extract intricate-patterns from EEG  and other similar bio-data sources.

 

Advanced EEG Analysis Techniques:

• Multitaper Spectral Analysis: This method offers high-resolution depictions of brainwave-activity, providing more-informative characterizations than traditional approaches. It's particularly effective in analyzing complex states like sleep.

Science Daily

• Wavelet Transform Analysis: Wavelet-analysis allows for the decomposition of EEG signals into time-frequency representations, facilitating the discrimination of complex and natural grasp types in brain-computer-interface-applications.

ArXiv

• Tensor-Based Multi-Dimensional-Frameworks: These frameworks enable holistic representation and analysis of EEG-data across multiple-dimensions, such as time, electrode-space, and frequency-bands, enhancing the accuracy and efficiency of EEG-signal-processing.

ArXiv

• Machine-Learning and Deep-Learning-Algorithms: Artificial-intelligence techniques are employed to decode intricate brain-states, propelling neuroscience into a future with many possibilities for understanding and interpreting EEG-data.

Springer Link

Leading Research Institutions:

• Center for Neuro-Engineering & Cognitive Science, University of Houston: This center focuses on developing advanced digital-signal-analysis tools for single-trial event-related-potential (EP) analysis, aiming to improve time-resolution in EEG-studies.

CNECS

• Laboratory of Brain, Behavior, and Pharmacology, UCLA: Researchers here have developed the cordance method, a quantitative-EEG-technique that combines absolute and relative power measures to assess regional-brain-activity, with applications in studying neurological and psychiatric disorders.

Wikipedia

• Computational Neurobiology Laboratory, Salk Institute: This lab has contributed to the development of tools like EEGLAB, a MATLAB toolbox for processing EEG and other electrophysiological-signals, implementing independent-component-analysis (ICA), time/frequency-analysis, and artifact-rejection.

Wikipedia

These institutions, among others, are at the forefront of EEG research, continually advancing methodologies to deepen our understanding of the brain's electrical activity and its implications for neurological health.

 

The 8 Best Sleep Trackers of 2024, Tested and Selected by Our Sleep Experts

Sleep-tracking devices can help you understand your sleeping patterns and stay healthier in the long-run. These are the best sleep-trackers on the market.

Caroline Igo,  Taylor Leamey - Dec. 21, 2024

Whoop 4.0

Best overall sleep tracker

$239 at Amazon

 

Oura Ring Gen3

Best wearable sleep tracker

$349 at Amazon

 

Investing in a sleep-tracker can help if you've been wanting to sleep better at night. Sleep-tracker data can be a reliable tool to understand how you're resting. These gadgets come in different shapes and sizes, from bulky watches to thin rings, and offer a range of different features you may or may not need. The best sleep trackers measure sleep phases, sleep quality, total amount of sleep and sleep disturbances, giving you a window into your overall sleep health. These also make great tech gifts for the holidays.

The best sleep tracker for your needs depends on your preferences, what you want to track and whether or not you'd like to wear the device. Our sleep team has tested 14 sleep-trackers over the years to bring you the best on the market. Here are the best smart rings, watches and mattress covers that track sleep performance. 

What is the best overall sleep-tracker? Our favorite overall sleep-tracker is the Whoop 4.0 Band, thanks to its in-depth sleep information and personalized recommendations. With no buttons or screens on the band, Whoop looks simple at first glance. However, the band's 5 LEDs, body-temperature sensors and photodiodes provide valuable insight into your sleep. A few of our CNET sleep experts have been using Whoop for months and have found it to provide the most-reliable and accurate sleep-data.

CNET's sleep experts say the Whoop provides reliable sleep data.

Aly Lopez/CNET

Whoop 4.0 Band$239 at Amazon

 

Taylor Leamey, senior sleep writer, tested the AI Whoop Coach and found it intuitive and a great feature if you're already interested in the band. You can ask the AI coach anything, and it will give you personalized answers and insights based on your sleep data. With Whoop, you get sleep data and assistance with interpreting everything.

Best Sleep Trackers for 2024

Aly Lopez/CNET

Whoop 4.0

Best overall sleep tracker

 

$239 at Amazon$239 at Walmart

 

Our take: The Whoop 4.0 sleep-tracker is for those who want to track both their sleep and their exercise. It’s for anyone who wants personalized sleep-insights. It has an integrated coach that provides personalized-recommendations, like when to go to bed or mindful breathing techniques, based on your sleep data. Whoop also is for those who don't want a screen to distract them at night. 

The review: "Of all the sleep trackers out there, Whoop gives me the most in-depth data on my sleep, strain and recovery,” said sleep writer Taylor Leamey. With Whoop’s advanced AI-powered technology and comprehensive data analysis, it is the ultimate sleep-tracker. Whoop provides excellent insights and monitors key sleep metrics like sleep-stages, heart-rate-variability, respiratory-rate and hours of sleep and restorative sleep, helping you better understand your sleep-patterns. It's one of the most-accurate we've tested, and the battery lasts up to 4-to-5-days without charging. 

Pros:

• Personalized recommendations based on your sleep data
• Provides reliable sleep data
• Removable battery pack allows you to charge from your wrist
• No distracting screen

Cons: 

• Chunky design can be uncomfortable to sleep with
• No screen for those who want one
• Expensive subscription plan
• 12-month minimum commitment 

Additional details: 

• Wearable: Yes
• Metrics measured: HRV, resting heart rate, recovery, respiratory rate, blood oxygen, body temperature, exercise and sleep performance 
• Trial: One month
• Warranty: Lifetime
• Subscription: $30 monthly (12-month commitment), $199 family membership, $239 up front for 12 months or $399 for 24 months
• Price: $$$

Caroline Igo/CNET - Oura Ring Gen3 - Best wearable sleep-tracker

$349 at Amazon$349 at Best Buy$349 at Oura

Our take: Oura Ring is an alluring, comfortable option for those who want convenient health and sleep monitoring without the discomfort of a bulky wrist wearable. Especially if you need a high-tech, lightweight device to fit into your existing wardrobe, Oura’s sleek design is snug and available in two styles and six finishes. In addition to providing overall sleep-readings, Oura Ring is a great option for those who want insights into other aspects of health like stress, menstrual-cycles and illness-monitoring.

The review: I wear my Oura Ring every day, and I often forget I even have it on. It’s become a part of my daily routine. I wake-up and check my sleep-score, my readiness-score for the day and cycle insights. I love that it automatically tracks activity and can be integrated with other apps such as Apple Health, Strava and Google Fit. It has excellent battery-life -- Oura claims it lasts up to 7-days, but I’ve found it's more like 5-to-6 days, which is still great. Over time, the Oura Ring gets to know your body and provides highly accurate insights and customized reports to help you understand your overall health and well-being. 

Pros:

• Comfortable, sleek ring design 
• Longer battery life compared to watch trackers
• Sleep score makes understanding sleep-quality easy
• Personalized health data and recommendations 

Cons:

• $6 monthly subscription is required for most features
• Depending on the finger and hand you wear it on, the ring can easily get scratched
• Doesn’t have snore detection 

Additional details: 

• Wearable: Yes
• Metrics measured: HRV, resting heart rate, respiratory rate, blood oxygen level, body temperature, stress, activity, menstrual cycle and sleep performance
• Trial: One month 
• Warranty: 1-year warranty
• Subscription: $6 per month
• Price: $$$

 

Now for a review of the formal scientific comparison I read back in 2022 and purchased an upgraded device, an Apple watch and Apple phone to replace my previous FitBit.

 

Nat Sci Sleep. 2022 Mar 22;14:493–516. doi: 10.2147/NSS.S348795

Performance of Four Commercial Wearable Sleep-Tracking Devices Tested Under Unrestricted Conditions at Home in Healthy Young Adults

Evan D Chinoy ^1,2, Joseph A Cuellar ^1,2, Jason T Jameson ^1,2, Rachel R Markwald ^1,✉

PMCID: PMC8957400  PMID: 35345630

Abstract

Purpose Commercial wearable sleep-tracking devices are growing in popularity and in recent studies have performed well against gold-standard sleep-measurement techniques. However, most studies were conducted in controlled laboratory conditions. We therefore aimed to test the performance of devices under naturalistic unrestricted home-sleep conditions.

Participants and Methods Healthy young adults (n = 21; 12 women, 9 men; 29.0 ± 5.0 years, mean ± SD) slept at home under unrestricted conditions for 1-week using a set of commercial wearable sleep-tracking devices and completed daily sleep diaries. Devices included the Fatigue Science Readiband, Fitbit Inspire HR, Oura ring, and Polar Vantage V Titan. Participants also wore a research-grade actigraphy watch (Philips Respironics Actiwatch-2) for comparison. To assess performance, all devices were compared with a high-performing mobile sleep-electroencephalography headband-device (Dreem-2). Analyses included epoch-by-epoch and sleep-summary agreement comparisons.

Results Devices accurately tracked sleep-wake summary metrics (ie, time in bed, total sleep time, sleep efficiency, sleep latency, wake after sleep onset) on most nights but performed best on nights with higher sleep efficiency. Epoch-by-epoch sensitivity (for sleep) and specificity (for wake), respectively, were as follows: Actiwatch (0.95, 0.35), Fatigue Science (0.94, 0.40), Fitbit (0.93, 0.45), Oura (0.94, 0.41), and Polar (0.96, 0.35). Sleep stage-tracking performance was mixed, with high variability.

Conclusion As in previous studies, all devices were better at detecting sleep than wake, and most devices compared favorably to actigraphy in wake-detection. Devices performed best on nights with more-consolidated sleep-patterns. Unrestricted sleep TIB differences were accurately tracked on most nights. High-variability in sleep stage-tracking-performance suggests that these devices, in their current form, are still best utilized for tracking sleep-wake-outcomes and not sleep-stages. Most commercial-wearables exhibited promising performance for tracking sleep-wake in real-world conditions, further supporting their consideration as an alternative to actigraphy.

Keywords: validation, actigraphy, mobile-EEG, consumer-sleep-technologies, habitual-sleep

Introduction Over the past decade, commercial sleep-tracking devices have become increasingly advanced in their technology and ubiquitous in their use across the population. Millions of new devices are sold each year, and estimates indicate that demand is increasing as the consumer sleep technology market grows rapidly into the billions of dollars.1 Beyond individual consumers, the sleep field has taken much interest in sleep-tracking technologies due to their potential for offering a relatively low-burden, low-cost, and reliable real-time sleep-tracking capability compared with standard sleep measurement techniques like polysomnography (PSG) and actigraphy. Major professional societies such as the American Academy of Sleep Medicine and Sleep Research Society have taken much interest in shaping the conversation around the present and future of sleep-tracking technologies – e.g., how to evaluate device performance,2 and determining guidelines for whether to potentially use data from commercial sleep-tracking technologies in clinical sleep medicine practice.3,4 

Encouragingly, many recent primary studies5–14 and reviews2,15–18 evaluating the performance of the newest commercial devices have demonstrated their improved sleep-tracking performance against many of the earliest device models released on the consumer market ~5-10 years ago.19–22 Several studies have even found that, compared with the gold-standard PSG, some commercial devices perform as well as or better than the current standard mobile sleep measurement methodology of research-grade actigraphy.6,7,10,11,13

Devices worn on the wrist (as a watch) or on the finger (as a ring) are some of the most popular types of wearables, as they offer the ability to track behavioral and physiological data from a convenient and comfortable location with virtually no requirements for prior training, time for setup, or need for manual input (i.e., most devices track sleep automatically and passively and can be continuously used over several nights on a single battery charge).23–25 Although many popular sleep-tracking devices have been subjected to initial performance evaluations by researchers, the pace of such research lags behind the more rapid release of new device models onto the market each year. Thus, with the increasing demands from consumers and sleep researchers and clinicians for an accurate sleep-tracking capability, additional studies are needed to evaluate which wearable device models are best and what level of reliability to expect from their performance.24

Most device performance studies have been limited to healthy participants in controlled environments (eg, sleep lab) under fixed time in bed (TIB) conditions. Those types of studies are a necessary first step in the recommended device validation testing process.2,26–28 The next stage of testing is to confirm that the findings obtained under ideal conditions will generalize to the real-world conditions in which those devices are most often used. To this end, we aimed to evaluate the sleep-tracking validation performance of four commercial wearable devices under unrestricted home sleep conditions in healthy young adults.

Participants and Methods

Participants A total of 21 healthy young adults (12 women, 9 men; 29.0 ± 5.0 years, mean ± SD) participated. Screening consisted of a self-report medical history questionnaire that assessed the following exclusion criteria: age <18 or >40 years, body mass index <18.5 or ≥30.0 kg/m² (25.2 ± 2.9, mean ± SD), any diagnosed sleep, mental health, or other medical disorder, use of any illegal drugs or sleep medications (over-the-counter or prescription) in the previous month, current pregnancy, any physical or living condition affecting the ability to sleep uninterrupted, and any night shift work or travel >1 time zone within the previous two weeks.

The study protocol was approved by the Naval Health Research Center Institutional Review Board and was conducted in accordance with the Declaration of Helsinki. Participants provided informed consent prior to the study and were compensated with gift cards.

Study Protocol Participants slept at home with a set of commercial sleep-tracking devices for 1 week. Sleep schedules were unrestricted (ie, participants could self-select to sleep at any time and duration over the study week). Though daytime sleep episodes and naps were allowed under these criteria, according to the sleep diaries and device data no participants chose to nap or sleep during the daytime. Thus, all sleep episodes evaluated in this study occurred at night. All devices were worn simultaneously during all sleep episodes. Within one hour after waking up, participants reported their bed and wake times using a digital sleep diary (based on the consensus sleep diary29) programmed on the data capture and management system app Smartabase (Fusion Sport; Milton, QLD, Australia) using a tablet computer (iPad; Apple Inc.; Cupertino, CA, USA). When not sleeping, participants were required to wear the set of devices as much as possible but could take devices off if needed at specified times, such as when needing to charge the device battery, when device(s) would interfere with their work duties, or while showering or engaged in other activities where devices could be submerged or damaged. Participants were instructed to report these device removal times using the Smartabase app.

Participants were allowed to engage in their habitual behaviors during the study week, including exercise and caffeine intake, and were required to report on these behaviors daily by using the Smartabase app. However, alcohol intake was not allowed during the study week. Researchers completed daily compliance checks online with the Smartabase app to ensure that participants were syncing the devices and that they completed their sleep diaries and other logs daily.

Commercial Sleep-Tracking Devices Tested Four commercial sleep-tracking wearable devices were tested in the study: Fatigue Science Readiband (version 5; Fatigue Science; Vancouver, BC, Canada), Fitbit Inspire HR (Fitbit, Inc.; San Francisco, CA, USA), Oura ring (2nd generation; ŌURA Health Oy; Oulu, Finland), and Polar Vantage V Titan (Polar Electro Oy; Kempele, Finland). On the first study day, researchers confirmed that all devices could be worn comfortably and in accordance with each device’s instructions. The four wrist devices (including the actigraphy watch, described below) were worn in pairs on each wrist. The actigraphy watch and Polar Vantage V Titan were worn as a pair on one wrist (with the Polar always worn most proximal to the wrist), and the Fatigue Science Readiband and Fitbit Inspire HR were worn as a pair on the other wrist (with the Fitbit always worn most proximal to the wrist). Wrist placement for each pair of devices was counterbalanced between participants, with half assigned to wear each pair on either their dominant or non-dominant wrist at the beginning of the study. During the middle of the study (after the fourth study night), participants were instructed to switch the device pairs to the other wrist, thus allowing an approximately equal number of assessment nights for either the dominant or non-dominant wrist. Participants wore the Oura ring on their non-dominant ring finger during the entire study week. See Supplementary Materials for details on the device software and firmware versions.

Participants were instructed to sync the data from all commercial devices using their respective apps on the tablet each day, within an hour of waking up. All commercial devices tested had photoplethysmography (PPG) heart rate sensors except the Fatigue Science Readiband. Thus, in addition to the standard sleep-wake outputs, the sleep algorithms for the three devices with PPG heart rate sensors (Fitbit Inspire HR, Oura ring, and Polar Vantage V Titan) also included outputs for individual sleep stage classifications of light, deep, and rapid eye movement (REM) sleep.

Comparison of Standard Sleep Methodologies

Actigraphy To assess sleep-wake data with actigraphy, participants wore the Actiwatch-2 (Philips Respironics, Inc.; Murrysville, PA, USA), a research-grade wrist actigraphy watch device (hereafter referred to as “Actiwatch”). Like the commercial devices, participants wore the Actiwatch during all sleep episodes and during as much of their waking hours as possible. Actiwatch sleep and wake data were analyzed in 30-sec epochs using the medium sensitivity threshold with the software package Actiware (version 6.0.9; Philips Respironics, Inc.; Murrysville, PA, USA). The medium sensitivity threshold is recommended for actigraphy analysis with healthy young adults; therefore, the medium threshold results are the primary Actiwatch results presented in the main text. For comparison, Actiwatch results were also analyzed using the low and high sensitivity threshold settings in Actiware and are presented in the Supplemental Materials. The Actiwatch data were post-processed to analyze each sleep episode using the bed and wake times reported in the participants’ daily sleep diaries.

Mobile EEG Headband The Dreem-2 (research version; Dreem; Paris, France), a high-performing wireless multi-channel dry mobile sleep electroencephalography (EEG) headband,30 was used as the primary reference standard of comparison to actigraphy and the commercial devices on sleep-wake and sleep stage outcomes (hereafter referred to as “Dreem”). In a performance validation study conducted by the company,30 Dreem was evaluated against PSG in a sample of healthy adults using the consensus scoring of 5 certified and experienced PSG scorers. Dreem was found to have high concordance for raw signal feature detection and relative EEG spectral band power with PSG, as well as low mean biases on sleep-wake summary outcomes, 83.5% overall scoring accuracy with PSG across 5 stages (i.e., all sleep stages and wake), and 74.0% accuracy with PSG for wake-epochs alone (i.e., specificity).

Comparatively, the consensus PSG scoring-accuracy between the 5 scorers was 86.4% for 5-stage scoring and 82.5% for wake-epochs; thus, the level of performance achieved with the Dreem compared with PSG was only slightly below the agreement among the experienced-PSG-scorers. Based on its demonstrated high-performance with a similar demographic of healthy-adults and its beneficial-features for multi-night-home-sleep-monitoring (e.g., good-comfort, ease-of-use, and ability to remotely-export-data each-day), we determined that Dreem was a practical and appropriate alternative to PSG for use in the current multi-night home-sleep-study as the reference-standard-method for sleep-scoring-classification.

Dreem data were collected in 30-sec epochs from the Dreem app’s sleep-scoring-algorithm. The Dreem provides sleep-stage-scores of N1, N2, N3, and REM. For standardization of the non-REM-stages to the commercial device stage classifications, N1 and N2 from Dreem were combined to form the light-sleep-stage and N3 was classified as deep-sleep.

For each sleep-episode, participants were instructed to: (1) put on the Dreem-headband a few minutes before getting into bed; (2) carry-out the required calibration-procedures and impedance-check for signal-quality using the Dreem-app on the tablet; and (3) initiate the sleep-episode-recording manually, using the app at their self-selected-bedtime. Participants were instructed to start the recording only at the top of a new clock-minute (i.e., zero-seconds after the minute using the tablet’s clock for reference). After their final-awakening in the morning, participants removed the Dreem-headband and wirelessly-synced the data with the app on the tablet. Unlike the Actiwatch and the commercial devices being tested, the Dreem was only worn during sleep-episodes, and the participants were instructed to charge the headband’s battery between nights.

Device Data Export Procedures Device data were exported from online-portals set-up by the device companies for researchers to manage participant accounts: Readi for the Fatigue Science Readiband, Oura Teams for the Oura ring, and Polar AccessLink API for the Polar Vantage V Titan. Fitbit, Inc. does not directly provide device data exports to researchers via the standard app or website accounts, therefore Fitbit Inspire HR data were exported via Fitabase (Small Steps Labs, LLC; San Diego, CA, USA), a licensed third-party-data-management-platform.

Missing-Data Procedures All participants completed the 7 data collection nights, except for one who withdrew for a personal reason after completing 6 nights in the study. Therefore, across the 21 participants there were collectively 146 potential nights of sleep-data. However, occasional issues occurred that affected the total amount of data included in the analyses.

For the commercial-devices, the number and percentage of nights from each device that had no recorded sleep episode data were as follows: Fatigue Science Readiband (1, 0.7%), Fitbit Inspire HR (0, 0.0%), Oura ring (3, 2.1%), and Polar Vantage V Titan (5, 3.4%). Additionally, a few nights recorded by the Fitbit and Polar devices (12 and 11 nights [8.2% and 7.8%], respectively) had poor-signal-quality (likely from the PPG-heart-rate-sensor) which caused those devices to revert to their binary sleep-wake algorithms instead of the current multi-sensor-algorithms that output sleep-stages. Those nights were removed from the final analyses of each device for the sleep-wake and sleep-stage outcomes.

The Actiwatch recorded continuously over the study week, so all nights had usable sleep-wake Actiwatch data for analysis. The Dreem recordings were manually started and stopped by the participants on the Dreem app each night. Thus, a few issues occurred that caused missing Dreem data for the whole night or part of the night. In some cases, a night of data was not recorded or saved because of a malfunction with the Dreem headband or app, or due to participant user error (10 nights, 6.8%). In other cases, the headband EEG leads may have been accidentally displaced from shifting during part of the night and may have produced a poor quality signal, which caused the Dreem algorithm to score some epochs during the TIB as movement time (MT) (30 nights, 22.1%). When a night contained a total of >10 min scored as MT, the Dreem data for that night were removed from the final sleep summary analyses (20 nights, 14.7%). An additional 4 nights (2.9%) were excluded due to participant user or reporting errors resulting in the reported sleep diary TIB being >10 min longer than the Dreem recording duration.

Sleep-summary data were included in the final analyses only when, according to the criteria listed above, there was a complete pair of usable data on the same night from both the device and the Dreem. Likewise, epoch-by-epoch (EBE) data were analyzed for only the included nights over the time-aligned epochs for the device and Dreem, and when they both had valid scores (e.g., not an epoch of MT from the Dreem). If a device recorded sleep data outside the sleep diary’s TIB on a given night, the Dreem was not recording during those times and thus those device epochs did not contribute to the EBE analyses. Likewise, if a device started recording after the Dreem was started at bedtime (or if the device recording ended before the Dreem was ended at wake time), only the epochs while both the device and Dreem were concurrently being recorded were utilized for the EBE analyses. Epochs from each device and the Dreem were aligned by clock-time according to the closest temporal alignment.

Statistical Analysis Statistical outcomes followed the recommended standards for sleep-tracking device evaluation2 and were based on the open source R code of Menghini et al,28 with some minor customizations. The sample sizes for the sleep summary and EBE analyses were dictated by the availability of valid pairs of data between each device and reference. The final sleep summary sample sizes are shown in the respective tables for each outcome, and the total EBE sample sizes are shown in the EBE contingency tables (Table S1) in the Supplementary Materials.

The following sleep summary measures were calculated for the reference (Dreem or sleep-diary), the Actiwatch, and each commercial device on each night: TIB (the total duration of a recorded sleep episode), total sleep time (TST; total duration of all sleep), sleep efficiency (SE; percentage of TST over the TIB), sleep onset latency (SOL; time taken to initially fall asleep), and wake after sleep onset (WASO; time awake in each sleep episode occurring after SOL).

For the three devices that classify sleep-stages, the total duration of light, deep, and REM sleep-stages on each night were also compared with the Dreem. Summary sleep-wake and sleep stage data were taken from each device’s summary data export files; however, the Fitbit Inspire HR and Polar Vantage V Titan did not differentiate their wake-output between that at the beginning versus middle-of-the-night, therefore SOL and WASO for those two devices were calculated from the EBE data exports instead. TIB for the Actiwatch was set to the TIB reported in the sleep-diary. Therefore, the Actiwatch TIB did not vary from the Dreem TIB and was not included in the TIB analyses.

For the sleep-summary-outcomes, Bland-Altman plots31 were generated to visualize discrepancies between the commercial-devices and the reference (Dreem or sleep-diary). In the Bland-Altman plots, differences in measurement were plotted against the mean of the two devices, along with reference lines that displayed the bias (the average difference between the device and reference) and the upper and lower 95% limits of agreement (two-standard-deviations from the bias). The plots, based on the Menghini et al28 framework, also permitted the visualization of proportional-bias (i.e., non-zero slope of the bias line, tested using linear regression methods that model the relationship between discrepancy size and mean measurement magnitude) and heteroscedasticity (i.e., changes in variability across the measurement scale, tested using linear-regression-methods that model the standard-deviation of the differences in terms of mean measurement size).28,32 P-values were considered statistically-significant for p<0.05.

The following EBE agreement statistics were calculated for the analysis of all sleep-versus-wake epochs for each device and the Actiwatch on each night, in comparison to the temporally-aligned epochs as scored by Dreem: sensitivity (true-positive-rate; the proportion of Dreem sleep-epochs that were correctly detected as sleep by the device), specificity (true-negative-rate; the proportion of Dreem wake-epochs that were correctly detected as wake by the device), positive-predictive-value (PPV; proportion of device-scored sleep epochs that were Dreem-scored sleep), negative-predictive-value (NPV; proportion of device-scored wake epochs that were Dreem-scored wake), accuracy (proportion of all device sleep and wake epochs correctly detected as Dreem-scored sleep or wake), and the prevalence-and-bias-adjusted-kappa (PABAK; Cohen’s-kappa-weighted to account for the amount of inequality between the number of sleep and wake epochs).

For the three devices that classify-sleep-stages, the EBE-agreement-statistics were also calculated for each stage versus the combination of all other classifications (e.g., EBE agreement for light sleep was calculated as the light sleep epochs versus the combination of all wake, deep, and REM epochs, etc.). Because there are multiple classification possibilities with sleep stage analysis, the proportions of misclassification were also determined for each stage (e.g., how often Dreem light sleep epochs were misclassified by a device as deep sleep, etc.). EBE results are presented in two ways: (1) all epochs, combining all individual epochs across all participants and nights, and (2) between nights, whereby EBE metrics were calculated for each night and then averaged across all nights (irrespective of participant).

Epoch durations for each device were either 30 or 60 sec. The Actiwatch, Fitbit Inspire HR, and Oura ring output sleep data in 30-sec epochs, which were directly matched with the Dreem which also outputs sleep data in 30-sec epochs. The Fatigue Science Readiband and Polar Vantage V Titan output sleep data in 60-sec epochs, which were divided into two 30-sec epochs and temporally aligned with the closest recorded Dreem epochs.

All preprocessing and statistical output were generated using the computing language R, version 4.0.5 (R Foundation, Vienna, Austria).

Results

Sleep-Wake Summary Agreement

Sleep-wake summary results are shown in Table 1 and corresponding Bland-Altman plots in Figures 1–5. Expanded sleep-wake summary results are presented in Supplementary Table S2.

Table 1

Sleep-Wake Summary Agreement

Outcome	Device	n	Device Mean (SD)	Reference Mean (SD)
TIB	Fatigue Science Readiband	112	459.0 (85.2)	437.4 (78.9)
	Fitbit Inspire HR	106	459.8 (78.6)	443.3 (73.6)
	Oura Ring	111	460.0 (90.3)	439.9 (74.0)
	Polar Vantage V Titan	103	452.0 (75.6)	445.5 (69.4)
TST	Actiwatch	113	406.1 (77.4)	401.3 (82.6)
	Fatigue Science Readiband	112	406.0 (89.4)	401.0 (82.9)
	Fitbit Inspire HR	106	406.4 (70.1)	406.6 (78.8)
	Oura Ring	111	412.5 (77.5)	403.1 (78.8)
	Polar Vantage V Titan	103	418.5 (74.0)	411.6 (72.1)
SE	Actiwatch	113	92.4 (3.2)	91.2 (6.9)
	Fatigue Science Readiband	112	88.0 (7.5)	91.3 (7.0)
	Fitbit Inspire HR	106	94.6 (2.9)	91.3 (7.0)
	Oura Ring	111	89.8 (4.9)	91.2 (7.0)
	Polar Vantage V Titan	103	92.5 (4.6)	92.2 (4.6)
SOL	Actiwatch	113	2.3 (1.7)	8.5 (5.6)
	Fatigue Science Readiband	112	18.8 (17.3)	8.5 (5.6)
	Fitbit Inspire HR	106	4.9 (3.4)	8.6 (5.6)
	Oura Ring	111	7.8 (5.5)	8.5 (5.6)
	Polar Vantage V Titan	103	0.6 (1.0)	8.5 (5.4)
WASO	Actiwatch	113	30.2 (12.7)	28.1 (22.5)
	Fatigue Science Readiband	112	21.2 (23.8)	27.9 (22.5)
	Fitbit Inspire HR	106	38.0 (13.1)	28.1 (21.9)
	Oura Ring	111	39.7 (32.0)	28.4 (22.5)
	Polar Vantage V Titan	103	25.8 (8.6)	25.4 (17.5)

Open in a new tab

Notes: Sleep-wake summary agreement between devices and reference. The sleep diary served as reference for TIB, and the Dreem headband served as reference for TST, SE, SOL, and WASO. All nights with available data for both the device and reference were included in analyses, with the total number of nights (n) indicated in each row. Mean and SD are shown for each device and reference.

Abbreviations: SD, standard deviation; SE, sleep efficiency; SOL, sleep onset latency; TIB, time in bed; TST, total sleep time; WASO, wake after sleep onset.

Figure 1

Open in a new tab

Bland-Altman plots: TIB.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for deviation in TIB for the devices compared with the reference sleep diary. Black circles are individual nights. Dashed lines represent the 95% confidence intervals around the bias and limits of agreement lines. Gray shaded regions on the right y-axis are density plots showing the distribution of individual night biases. Zero on the y-axis represents no difference, with positive and negative y-axis values indicating an overestimation or underestimation, respectively, compared with the reference. Diagonal mean bias lines indicate significant proportional bias. Non-parallel limits of agreement lines indicate significant heteroscedasticity.

Abbreviation: TIB, time in bed.

Figure 2

Open in a new tab

Bland-Altman plots: TST.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for minutes of TST for the devices compared with the reference Dreem headband. See Figure 1 notes for additional figure details.

Abbreviation: TST, total sleep time.

Figure 3

Open in a new tab

Bland-Altman plots: SE.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for percentage of SE for the devices compared with the reference Dreem headband. See Figure 1 notes for additional figure details.

Abbreviation: SE, sleep efficiency.

Figure 4

Open in a new tab

Bland-Altman plots: SOL.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for minutes of SOL for the devices compared with the reference Dreem headband. See Figure 1 notes for additional figure details.

Abbreviation: SOL, sleep onset latency.

Figure 5

Open in a new tab

Bland-Altman plots: WASO.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for minutes of WASO for the devices compared with the reference Dreem headband. See Figure 1 notes for additional figure details.

Abbreviation: WASO, wake after sleep onset.

For TIB (Table 1 and Figure 1), most nights for each device clustered around the mean bias line, which itself deviated little from 0. However, in each device, a small subset of nights exhibited biases larger than an hour. Mean biases were all low, near zero, and no devices exhibited proportional bias.

TST mean biases for the Actiwatch and all commercial devices were close to the Dreem-determined TST as indicated in Table 1 and Figure 2, with the majority of individual nights showing low bias (eg, <30 min) although some nights had larger biases. The individual night data indicate that no device had consistent under or over-estimation biases in TST, though the Actiwatch and Fitbit Inspire HR each had significant negative proportional mean biases (ie, mean bias decreased as TST increased).

The SE summary results (Table 1 and Figure 3) were similar to the TST results in that the majority of nights for the Actiwatch and all devices exhibited low SE biases versus Dreem. A clear pattern emerged across all devices of reduced variability as SE increased. Additionally, the Actiwatch, Fitbit Inspire HR, and Oura ring each exhibited significant negative proportional mean biases.

SOL (Table 1 and Figure 4) mean biases were low for all devices, and all except the Oura ring exhibited significant proportional bias (positive for Fatigue Science Readiband and negative for Actiwatch, Fitbit Inspire HR, and Polar Vantage V Titan). Individual night biases for all devices were lowest when SOL was also low. WASO (Table 1 and Figure 5) biases showed more variability but were also lowest when WASO was low. Proportional mean biases were significant for all devices except the Fatigue Science Readiband, with negative proportional biases for Actiwatch, Fitbit Inspire HR, and Polar Vantage V Titan and positive proportional bias for Oura ring.

Sleep Stage Summary Agreement

Sleep stage summary results are shown in Table 2 and corresponding Bland-Altman plots in Figures 6–8. Expanded sleep stage summary results are presented in Supplementary Table S3.

Table 2

Sleep Stage Summary Agreement

Stage	Device	n	Device Mean (SD)	Reference Mean (SD)
Light	Fitbit Inspire HR	106	243.5 (53.0)	189.4 (54.3)
	Oura Ring	111	210.8 (66.0)	189.0 (53.9)
	Polar Vantage V Titan	103	226.1 (90.5)	190.7 (52.9)
Deep	Fitbit Inspire HR	106	70.1 (21.9)	80.1 (29.3)
	Oura Ring	111	124.1 (41.7)	81.4 (29.2)
	Polar Vantage V Titan	103	104.5 (92.1)	82.8 (28.6)
REM	Fitbit Inspire HR	106	92.8 (30.2)	137.2 (54.8)
	Oura Ring	111	77.5 (34.8)	132.7 (52.6)
	Polar Vantage V Titan	103	87.8 (40.7)	138.2 (54.4)

Open in a new tab

Notes: Sleep stage summary agreement between devices and reference. The Dreem headband served as reference for all sleep stage outcomes. Results are shown for all devices that output sleep stage classifications. See Table 1 notes for additional table details.

Abbreviations: REM, rapid eye movement; SD, standard deviation.

Figure 6

Open in a new tab

Bland-Altman plots: light sleep.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for minutes of light sleep for the devices compared with the reference Dreem headband. See Figure 1 notes for additional figure details.

Figure 7

Open in a new tab

Bland-Altman plots: deep sleep.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for minutes of deep-sleep for the devices compared with the reference Dreem-headband. See Figure 1 notes for additional figure details.

Figure 8

Open in a new tab

Bland-Altman plots: REM sleep.

Notes: Plots depict the mean bias (solid red line) and upper and lower limits of agreement (solid gray lines) for minutes of REM sleep for the devices compared with the reference Dreem headband. See Figure 1 notes for additional figure details.

Abbreviation: REM, rapid-eye-movement.

In general, for the three devices that provide sleep-stage-classifications, the summary results and individual night biases for sleep-stages were mixed and more-variable than the sleep-wake outcomes. Light sleep (Table 2 and Figure 6) biases were variable and generally overestimated compared with Dreem, and there were significant positive proportional mean biases for the Oura ring and Polar Vantage V Titan. Deep sleep (Table 2 and Figure 7) biases were also highly variable for most nights, with Oura ring and Polar Vantage V Titan producing overestimates on many nights. Of note, the Polar Vantage V Titan had major overestimations on 10 nights (spread among 7 participants) due to it classifying all sleep as deep sleep only. Each device exhibited significant proportional mean biases for deep sleep, which was negative for the Fitbit Inspire HR but positive for the Oura ring and Polar Vantage V Titan. REM sleep (Table 2 and Figure 8) biases were also highly variable for each device, with the majority of nights producing underestimates versus Dreem and significant negative proportional mean biases.

EBE Sleep-Wake Agreement EBE agreement outcomes for sleep versus wake epochs for each device were compared with Dreem and are shown in Table 3. Values were calculated in two ways: (1) All epochs combined across participants and nights, and (2) Between nights. Both calculations yielded similar means. For all epochs, sensitivity levels for Actiwatch and each device were very high with a small range (0.93–0.96), while specificity levels were all low to medium (0.35–0.45). Notably, the specificity level for each commercial device was higher than Actiwatch, except for the Polar Vantage V Titan which was similar to Actiwatch. To a large extent, the other EBE outcomes reflect the sensitivity and specificity results. Like sensitivity, accuracy and PPV values across devices were all high and in a narrow range, while NPV was low to medium like specificity. PABAK values were also similar with a narrow range of values across devices.

Table 3

EBE Sleep-Wake Agreement

Device	Sensitivity		Specificity		Accuracy		PPV		NPV		PABAK
	All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights
Actiwatch	0.95	0.95 (0.02) [0.95, 0.96]	0.35	0.37 (0.15) [0.34, 0.4]	0.90	0.90 (0.06) [0.89, 0.91]	0.94	0.94 (0.06) [0.93, 0.95]	0.38	0.39 (0.16) [0.36, 0.42]	0.80	0.80 (0.11) [0.78, 0.82]
Fatigue Science Readiband	0.94	0.94 (0.07) [0.92, 0.95]	0.40	0.43 (0.20) [0.39, 0.46]	0.90	0.89 (0.07) [0.88, 0.90]	0.95	0.94 (0.06) [0.93, 0.95]	0.37	0.45 (0.22) [0.41, 0.49]	0.79	0.78 (0.15) [0.75, 0.81]
Fitbit Inspire HR	0.93	0.93 (0.03) [0.93, 0.94]	0.45	0.50 (0.17) [0.47, 0.54]	0.89	0.89 (0.05) [0.88, 0.90]	0.95	0.95 (0.06) [0.94, 0.96]	0.36	0.36 (0.15) [0.33, 0.39]	0.79	0.78 (0.11) [0.76, 0.81]
Oura Ring	0.94	0.94 (0.03) [0.94, 0.95]	0.41	0.44 (0.17) [0.41, 0.47]	0.90	0.90 (0.06) [0.89, 0.91]	0.95	0.95 (0.06) [0.94, 0.96]	0.37	0.39 (0.16) [0.36, 0.42]	0.81	0.80 (0.11) [0.78, 0.82]
Polar Vantage V Titan	0.96	0.96 (0.02) [0.95, 0.96]	0.35	0.39 (0.16) [0.36, 0.42]	0.92	0.92 (0.04) [0.91, 0.93]	0.96	0.96 (0.04) [0.95, 0.97]	0.36	0.36 (0.15) [0.33, 0.38]	0.84	0.84 (0.08) [0.82, 0.85]

Open in a new tab

Notes: Proportions for EBE agreement metrics are shown for all sleep epochs (versus wake epochs) for all combined epochs as well as between all nights (irrespective of participant) for the devices compared with the corresponding epochs from the reference Dreem headband. Values for between-night analyses are presented as mean (SD) [95% confidence interval]. Higher values (closer to 1.0) indicate better performance on that metric.

Abbreviations: EBE, epoch by epoch; NPV, negative predictive value; PABAK, prevalence and bias adjusted kappa; PPV, positive predictive value; SD, standard deviation.

EBE Sleep Stage Agreement EBE agreement for each sleep stage are shown in Table 4. Compared with the sleep-wake EBE results, the sleep stage EBE performance levels were mixed. For light sleep, sensitivity and specificity were at a medium level for all devices. For deep and REM sleep, specificity levels were high while sensitivity levels were low to medium, indicating that devices are generally better at detecting when a given epoch is a classification other than deep or REM sleep than when it is that stage. Values were within a narrow range across devices within all the other stage outcomes. Accuracy for each device was at a medium level for light sleep, but at medium to high levels for deep and REM. PPV levels were mostly at a medium level while NPVs were higher, especially for deep and REM. PABAK levels for light sleep were low for each device, but medium for deep and REM.

Table 4

EBE Sleep Stage Agreement

Stage	Device	Sensitivity		Specificity		Accuracy		PPV		NPV		PABAK
		All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights	All Epochs	Between Nights
Light	Fitbit Inspire HR	0.71	0.71 (0.09) [0.70, 0.73]	0.59	0.60 (0.10) [0.58, 0.62]	0.64	0.65 (0.07) [0.63, 0.66]	0.57	0.57 (0.13) [0.54, 0.60]	0.73	0.73 (0.10) [0.71, 0.75]	0.29	0.29 (0.13) [0.26, 0.32]
	Oura Ring	0.61	0.60 (0.12) [0.58, 0.62]	0.64	0.64 (0.12) [0.62, 0.67]	0.62	0.62 (0.07) [0.61, 0.64]	0.56	0.56 (0.13) [0.54, 0.59]	0.69	0.68 (0.10) [0.66, 0.70]	0.25	0.25 (0.15) [0.22, 0.28]
	Polar Vantage V Titan	0.59	0.63 (0.10) [0.61, 0.65]	0.54	0.50 (0.10) [0.48, 0.52]	0.56	0.56 (0.07) [0.55, 0.57]	0.49	0.50 (0.13) [0.47, 0.52]	0.64	0.63 (0.11) [0.61, 0.66]	0.12	0.12 (0.14) [0.09, 0.15]
Deep	Fitbit Inspire HR	0.51	0.53 (0.23) [0.49, 0.58]	0.92	0.93 (0.04) [0.92, 0.94]	0.85	0.85 (0.06) [0.84, 0.86]	0.58	0.61 (0.23) [0.57, 0.66]	0.90	0.90 (0.06) [0.88, 0.91]	0.70	0.70 (0.11) [0.68, 0.72]
	Oura Ring	0.72	0.74 (0.22) [0.70, 0.78]	0.83	0.83 (0.10) [0.81, 0.85]	0.81	0.81 (0.07) [0.79, 0.82]	0.50	0.51 (0.17) [0.47, 0.54]	0.93	0.93 (0.06) [0.92, 0.94]	0.62	0.62 (0.14) [0.59, 0.64]
	Polar Vantage V Titan	0.45	0.46 (0.27) [0.41, 0.51]	0.82	0.81 (0.24) [0.76, 0.86]	0.75	0.75 (0.15) [0.72, 0.78]	0.37	0.41 (0.19) [0.37, 0.44]	0.86	0.87 (0.06) [0.86, 0.89]	0.51	0.50 (0.31) [0.44, 0.56]
REM	Fitbit Inspire HR	0.47	0.48 (0.14) [0.45, 0.51]	0.91	0.91 (0.05) [0.90, 0.92]	0.77	0.77 (0.08) [0.76, 0.79]	0.70	0.70 (0.16) [0.66, 0.73]	0.79	0.79 (0.11) [0.77, 0.82]	0.55	0.55 (0.16) [0.52, 0.58]
	Oura Ring	0.39	0.40 (0.17) [0.37, 0.43]	0.92	0.92 (0.05) [0.91, 0.93]	0.76	0.76 (0.09) [0.74, 0.78]	0.68	0.68 (0.19) [0.64, 0.71]	0.78	0.78 (0.11) [0.76, 0.80]	0.51	0.52 (0.18) [0.48, 0.55]
	Polar Vantage V Titan	0.36	0.40 (0.16) [0.37, 0.43]	0.88	0.87 (0.07) [0.86, 0.88]	0.71	0.71 (0.10) [0.69, 0.73]	0.59	0.57 (0.21) [0.53, 0.61]	0.75	0.75 (0.12) [0.73, 0.78]	0.43	0.43 (0.20) [0.39, 0.47]

Open in a new tab

Notes: Proportions for EBE agreement metrics are shown for each sleep stage (versus the combination of all other possible classifications) for all combined epochs as well as between all nights (irrespective of participant) for the devices compared with the corresponding epochs from the reference Dreem headband. Values for between-night analyses are presented as mean (SD) [95% confidence interval]. Higher values (closer to 1.0) indicate better performance on that metric. Results are shown for all devices that output sleep stage classifications.

Abbreviations: EBE, epoch by epoch; NPV, negative predictive value; PABAK, prevalence and bias adjusted kappa; PPV, positive predictive value; REM, rapid eye movement; SD, standard deviation.

EBE Sleep Stage Misclassification Errors EBE sleep stage misclassifications were calculated in two ways: (1) Error rates for each device misclassifying each stage, given what the Dreem scored (Table 5A), and (2) Classification differences of what the Dreem scored, given what the device scored (Table 5B). The largest source of error for either error calculation was with light sleep epochs. Among wake, deep, and REM epochs as scored by the Dreem, the devices misclassified those epochs as light sleep between 24–47% of the time. Additionally, wake, deep, and REM epochs as scored by devices were misclassified and, according to the Dreem, should have been scored as light sleep 22–44% of the time. Epochs scored as light by devices were misclassified and should have been scored as REM 24–29% of the time. Another substantial misclassification occurred when devices scored REM sleep, according to the Dreem those epochs should have been scored as wake 28–30% of the time.

Table 5

EBE Sleep Stage Misclassification Errors

A
Device	Wake (Dreem)			Light (Dreem)			Deep (Dreem)			REM (Dreem)
	Light (Device)	Deep (Device)	REM (Device)	Wake (Device)	Deep (Device)	REM (Device)	Wake (Device)	Light (Device)	REM (Device)	Wake (Device)	Light (Device)	Deep (Device)
Fitbit Inspire HR	0.37	0.04	0.14	0.07	0.11	0.10	0.03	0.42	0.04	0.09	0.41	0.03
Oura Ring	0.33	0.16	0.10	0.06	0.23	0.10	0.01	0.24	0.03	0.08	0.45	0.09
Polar Vantage V Titan	0.41	0.10	0.13	0.04	0.23	0.14	0.02	0.45	0.08	0.06	0.47	0.11
B
Device	Wake (Device)			Light (Device)			Deep (Device)			REM (Device)
	Light (Dreem)	Deep (Dreem)	REM (Dreem)	Wake (Dreem)	Deep (Dreem)	REM (Dreem)	Wake (Dreem)	Light (Dreem)	REM (Dreem)	Wake (Dreem)	Light (Dreem)	Deep (Dreem)
Fitbit Inspire HR	0.32	0.06	0.28	0.05	0.14	0.24	0.02	0.31	0.07	0.05	0.22	0.03
Oura Ring	0.31	0.03	0.28	0.06	0.09	0.29	0.04	0.37	0.10	0.04	0.25	0.03
Polar Vantage V Titan	0.29	0.06	0.30	0.05	0.17	0.29	0.03	0.44	0.16	0.04	0.30	0.08

Open in a new tab

Notes: (A) Proportions for device EBE sleep stage misclassification errors versus those scored by the reference Dreem headband. Dreem-scored classifications are the larger column categories, with the three possible device-scored misclassifications under each category. (B) Proportions of sleep stage EBE differences in Dreem-scored classifications from the device-scored epochs. Device-scored classifications are the larger column categories, with the three possible Dreem-scored differences in classification under each category. Results are shown for all devices that output sleep stage classifications.

Abbreviations: EBE, epoch by epoch; REM, rapid eye movement.

Actiwatch Sensitivity Threshold Comparisons Sleep-wake summary and EBE results for the Actiwatch analyzed with the low and high sensitivity thresholds are described in the text of the Supplementary Materials, with accompanying data presented in Supplementary Tables S4-S6 and depicted with Bland-Altman plots in Supplementary Figure S1. In general, the different sensitivity threshold settings exhibited mixed and opposite results for the sleep-wake detection capability of the Actiwatch compared with the recommended medium setting and the commercial devices. The low setting generally increased the amount of wake and decreased the amount of sleep classified across sleep episodes, and the high setting did the opposite. This corresponded to performance tradeoffs, resulting in the Actiwatch being more sensitive to wake or sleep with the low and high settings, respectively.

Discussion In general, the four commercial wearable devices we tested had similar sleep-tracking performance during real-world unrestricted home sleep in healthy-young-adults. Like many prior studies, the commercial devices were better at detecting sleep than wake, indicated by the high-sensitivity and low-to-medium specificity EBE-levels. The specificity-levels for most commercial devices were comparable to research-grade-actigraphy, revealing a similar or possibly greater ability than actigraphy to detect wake. Sleep-wake summary outcomes including TIB, TST, SE, SOL, and WASO were accurately tracked with low-bias on most-nights, but were generally most-accurate on nights with higher-SE. A few nights still exhibited large biases for sleep summary outcomes, but the specific causes of such bias or variability remain to be determined. Sleep-stages had greater bias and variability than the binary sleep-wake outcomes, suggesting that the ability of devices to track individual sleep-stages is worse than their ability to track sleep-wake in general.

Taken together, our findings suggest that the commercial wearable devices predominantly exhibited promising performance for tracking nighttime sleep-wake patterns in real-world unrestricted home conditions with healthy-young-adults.

Many of the findings in the present study are similar to those in previous studies with these and other sleep-tracking devices. High-sensitivity and low-to-medium specificity were expected, as these are consistent findings among previous studies testing the agreement of EBE sleep-wake classifications with either research-grade actigraphy or commercial devices.2,7,27,33 With this expected weakness of the sleep-wake-algorithms to under-detect wake-epochs (and general acceptance of this limitation by researchers for mobile sleep assessments with actigraphy methods34), a useful interpretative benchmark for evaluating the performance of commercial devices is to consider their specificity-levels in relation to those of actigraphy (i.e., the Actiwatch).

In this case, all 4 commercial devices performed at levels that were comparable to research-grade actigraphy. That the Fatigue Science Readiband, Fitbit Inspire HR, and Oura-ring appeared to exceed the specificity of the Actiwatch (at the medium-sensitivity threshold-setting) by 5–10% is intriguing and in-line with previous studies, but this result should be replicated before arriving at strong conclusions about potential superior performance of devices across other populations and settings.

Between-night comparisons of specificity resulted in slightly higher specificity levels than when all epochs were analyzed, but this may be due in part to differences in TIB and/or the amount of wake-epochs between nights. The low and high sensitivity threshold-settings did result in altered EBE results for the Actiwatch, including higher-specificity when the low-setting was utilized. However, the low-setting also came with tradeoffs, such as slightly-reduced sensitivity and NPV and larger mean-biases for most of the sleep-wake-summary-outcomes. This result suggests that, depending on the population and setting, the different actigraphy analysis settings may be warranted in some studies, but that the decision may result in performance tradeoffs. Most recent studies of newer models from major commercial device companies have yielded similar results for tracking sleep-wake-summary-measures such as TST, SE, SOL, and WASO, with comparable levels of accuracy and low-bias on most-nights.

In contrast, novel findings from the current study address the accuracy of commercial device algorithms for tracking unrestricted TIB in healthy young adults. Most previous device performance studies utilized designs with controlled lab conditions and fixed TIB (or allowing only a small TIB range). With the weeklong unrestricted home sleep design, we were able to gather data with wider distributions of TIB and clock times. We largely found that TIB was tracked accurately on most nights, however there were still considerable biases on several nights for each device. TIB bias and variability from the commercial devices may in part be due to participant activity levels around bedtime and wake time. The commercial devices utilize proprietary algorithms that track sleep automatically (which is a beneficial feature for low-burden, real-time sleep tracking), but such automatic algorithms may be impacted by a participant’s sedentary activity. For example, relaxing at night while awake in bed could cause automatic algorithms to start their sleep recording earlier than one’s actual bedtime. Actigraphy analyses were set to the reported TIB, as is standard procedure,34 and thus actigraphy would be less affected by sedentary activity surrounding the sleep episode compared with the commercial devices. Despite relying on automatic algorithms to start and end sleep recordings, the commercial devices still performed well on most nights in tracking TIB and other summary outcomes. TIB is a critical outcome for tracking sleep, because an inaccurate TIB will expand or restrict the sleep recording window and have downstream effects on the other sleep-wake outcomes. Additional studies should be conducted to examine device performance under conditions that better reflect the real-world variability in sleep schedules, such as in shift workers and under operational contexts that include different TIB durations and patterns such as shorter naps or split sleep schedules.

The predecessor Fitbit model to the Inspire HR, the Alta HR, has been compared directly to PSG by our group7 and others,10,11 and it displayed low levels of bias on summary and EBE performance that was also favorable to actigraphy. One study35 also examined the Fitbit Alta HR in insomnia patients at home for one week, finding that it displayed similar sleep-tracking ability as actigraphy (but did not test it against PSG or mobile EEG). Studies testing other recent Fitbit models found the Charge 3 was high-performing when tested against PSG in adolescents,12 and the Ionic performed among the best out of nine sleep-tracking devices tested at home versus mobile EEG in adults.14 Fitbit device models have been the focus of most device performance studies in the sleep field (likely due to their early entry as a major company in the emerging wearables market). Over the past 5–10 years, Fitbit has consistently improved their technology (eg, added a PPG heart rate sensor in newer models) and their sleep-tracking algorithm performance.16 To our knowledge, the current study is the first to examine sleep in the newer Inspire HR model from Fitbit, which we found performs like the more widely-tested Alta HR model as well as other recent models and may still be considered among the best sleep trackers tested.

The Oura ring showed overall high performance across sleep-wake summary and EBE outcomes in the current study. This result is consistent with findings from other recent performance studies that evaluated the Oura ring versus either PSG5,6,8,36 or mobile EEG.14 In the study with similar methods comparing Oura ring with mobile EEG at home,14 the Oura ring (along with Fitbit Ionic, referenced earlier) performed the best of nine commercial sleep-tracking devices tested. The current study tested the 2nd generation version of the Oura ring, although studies with the 1st generation8,36 Oura ring found good sleep-tracking performance along with the more recent studies of the 2nd generation ring5,14 despite methodological and sample demographic differences that may have resulted in higher EBE specificity levels reported in the other studies of the 2nd generation Oura ring. Thus, the ring form factor represents an alternative for high-performing sleep tracking along with the much more common wrist-worn devices.

We previously evaluated the Fatigue Science Readiband versus lab-controlled PSG, and found that it performed similar to or better than actigraphy on most sleep-wake outcomes.7 One other study also evaluated the Readiband under conditions similar to the current study (ie, at home versus mobile EEG), and found it generally performed well on most outcomes and ranked around the middle among nine sleep-tracking devices that also included the more high-performing Fitbit Ionic and Oura ring.14 The Readiband is not marketed as a direct-to-consumer device, rather it is typically marketed as part of a fatigue management platform with athletic teams and operational or shift working groups. The limited marketing of the Readiband may be among the reasons it has received less attention from consumers and researchers, despite the results from this study and prior studies suggesting it is a viable option for sleep-wake tracking and with longer battery life (~30 days) compared with the other devices tested (~4-7 days).

In general, the Polar Vantage V Titan exhibited mixed performance, with specificity that was similar to actigraphy but not as high as the other commercial devices. To our knowledge, the Vantage V Titan model from Polar has not been subjected to previous sleep-tracking performance testing, so it was unknown how it would perform relative to the other devices or a reference methodology. However, in a study of six commercial wearable devices versus actigraphy, a previous Polar device model (the A360) performed relatively poorly in its sleep-tracking performance.37 And compared with mobile EEG at home, another previous Polar model (the A370) had middling sleep-tracking performance among nine devices tested.14 One other area in which we found that the Polar Vantage V Titan performed worse than the other devices was in sleep stage tracking. Although no device performed particularly well at tracking sleep stages, on 10 nights (comprising ~10% of all nights tested and spread among 7 participants) the Polar classified all sleep that occurred as deep sleep. That outcome drastically biased the Polar’s sleep stage data on those nights and was a unique phenomenon among the devices in this study, and to our knowledge has not been reported in previous device studies. With several other commercial devices having been more thoroughly evaluated and with superior performance, it may be advisable to wait until Polar sleep-tracking algorithms have been updated and devices more thoroughly evaluated before selecting Polar devices for research purposes.

Sleep stage-tracking performance was mixed and highly variable compared with the binary sleep-wake outcomes. These results were expected given similar findings from our previous work7 and from others8,10–12,14 that tested sleep-stage-tracking of various devices versus PSG. Most sensitivity levels for the three devices that output stage data were at a medium-level, indicating that a device can only classify the correct stage for a given epoch around half the time. Although specificity-levels for deep and REM sleep were high, there are three other possible classifications for a given epoch. Thus, a high-specificity in the case of a sleep-stage does not offer confidence in understanding how that epoch should be classified. Additionally, as in our prior study,7 we again found that light-sleep was the most common stage for error, with Dreem-scored wake, deep, and REM epochs having a considerable chance of mistakenly being misclassified as light-sleep (and vice-versa). The other stage classifications were not as often mistaken with each other, except that all the devices also showed a high-proportion for misclassifying wake as REM-sleep.

There may be room for future devices/algorithms to improve, though this may represent a type of ceiling-effect on the ability of current algorithms to correctly classify sleep-stages from wearables. Improved sleep stage-tracking will likely remain a challenge largely due to the established scoring rules for each stage being based primarily on EEG-waveforms and phasic-events38 that do not (or potentially cannot) correlate precisely with signals measured only from a device placed on the wrist or finger.

We believe this study has several notable strengths: (1) This is one of the first studies to test device performance with unrestricted home sleep schedules that reflect naturalistic variation in TIB, representing a step-forward in understanding how the performance of commercial sleep-tracking devices compares with the more commonly utilized lab-controlled study-designs; (2) Unlike most performance studies that test only a single sleep-tracking-device, testing multiple devices simultaneously supported a higher throughput of devices tested and with multiple form factors (i.e., wrist-worn and ring devices); (3) The weeklong study design using mobile-EEG allowed us to collect up to 7 nights of high-quality sleep-data per-participant, without the typical burdens placed on researchers and participants that are required with lab-controlled PSG procedures; (4) Testing commercial devices alongside a research-grade actigraph allowed a direct comparison with the mobile sleep assessment standard methodology of actigraphy and opens the results up to interpretation as to whether commercial devices may serve as an adequate future alternative to actigraphy in sleep research and medicine; (5) The testing procedures and analyses follow recently recommended standards2,28 for sleep-tracking-device-performance-evaluation; and (6) This study was an independent investigation, without potential conflicts from the device companies.

Additionally, notable limitations of this study include the following:

1 - Mobile-EEG provides high-quality sleep-data that are relatively easier to collect in home settings, but it is not as accurate as traditional gold-standard-PSG. Therefore, using the Dreem mobile-EEG-headband and its associated auto-scoring algorithm30 is a highly-practical alternative as a mobile standard sleep assessment methodology, but is less-precise than PSG in scoring sleep-wake and sleep-stages;

2 - Although most data were high-quality and usable in the analyses, we experienced occasional data-loss with the commercial devices and the Dreem-headband due to user-error, technological malfunctions, low-signal-quality, or other reasons. New technologies can be unpredictable and challenging to use without error, especially when used in remote-settings outside the lab where researchers are not directly present. Having standardized-methods and employing daily-data and compliance-checks helped ensure that data-collection-issues were minimized and resolved quickly;

3 – EBE-analyses relied on concurrent-pairs-of-epochs between a device and Dreem. Therefore, device epochs that were recorded outside the actual TIB from the sleep-diary/Dreem-recording (which were likely to be wake epochs) and Dreem-epochs recorded within the TIB when a device may not have been recording at the same time, were not able to be included in the EBE-analyses;

 4 - Wearable-devices with PPG-sensors may be prone to error in heart-rate-tracking-accuracy due to demographic (e.g., race, ethnicity) and physical-differences in skin-types (e.g., skin-tone, hair-follicle-density) between participants,39 which could have unknown effects on the sleep-tracking-outcomes for the devices that use PPG-sensors in this study. Such data were not collected in this study, however future studies should collect data on individual demographics and skin-types to explore potential effects of those factors on wearable-device-sleep-tracking-performance;

 5 - Like many previous device-performance-studies, the current study was restricted to healthy young adults without reported sleep disorders or other medical conditions. Thus, we observed relatively-low-variability in the amounts and patterns of wake within the sleep-episodes, limiting the interpretation for how devices may perform in populations with sleep-initiation or sleep-maintenance issues that exhibit higher-levels of SOL and WASO (and lower-SE);

6 - The consumer-market has a high-turnover-rate of devices that track-sleep. We chose to test device models from companies that are established in the wearables-market, and that are widely-used by the public or by operational and athletic communities that use such devices to track their health and behavior. It is possible that the specific device models used in the present study may not be available or supported by the device companies in a few years, however they are likely to be replaced by comparable device models using similar sensors and sleep-tracking algorithms. Thus, the results are most-relevant for the specific models/algorithms but are nonetheless informative to the performance of sleep-tracking-devices from these companies in general for years into the future; and

7 - The focus of the present study was to report the basic-performance-evaluation-metrics of the commercial-devices versus mobile-EEG and actigraphy. Further analyses between and within participants across nights as well as demographic-comparisons, leveraging the full strengths of the diversity of participants and multi-night-study-design, are being prepared for inclusion in future publications.

Conclusion These findings demonstrate that several commercial devices that track sleep in unrestricted home sleep conditions are promising alternatives to research-grade-actigraphy. As a first-step toward eventual validation, several of these devices (or very-similar device models) were previously tested primarily under lab-controlled conditions and exhibited good performance compared with the gold-standard PSG. The current study extends those findings by showing that sleep-tracking-performance is largely similar under unrestricted home-settings with naturalistic-variation in sleep-schedules, which further supports the potential for using commercial-devices as a high-quality alternative to actigraphy for mobile-sleep-wake-assessments.

Altering the sensitivity-analysis-thresholds may enhance the performance of actigraphy for certain performance domains, albeit with some performance tradeoffs, and researchers are advised to consider carefully the population and setting of their study when choosing device and analysis settings. As in prior studies, however, commercial-devices exhibited more-modest performance for sleep-stage-tracking and were prone to higher-levels of bias and variability compared with the sleep-wake-outcomes. The recent advancements in sensor-technology and ubiquitous use of wearable-devices that track-sleep and other relevant behavioral and physiological signals represents a promising future where personal-health-insights are gained with tools that possess several major strengths over current standard assessment methodologies (e.g., lower-burden, more-widely-available, more-comfortable, less-expensive, multi-functional, and multi-day tracking on a single charge).

Given the potential importance of commercial-wearables in sleep-research and medicine, future studies are warranted to continue to assess the performance of these and other devices in broader settings and within more diverse populations of interest (e.g., patients with sleep-disorders, shift-workers, and operational-populations).

Acknowledgments This research was funded by the Office of Naval Research, Code 34. The authors wish to thank the study participants.

Abbreviations EBE, epoch by epoch; EEG, electroencephalography; MT, movement time; NPV, negative predictive value; PABAK, prevalence and bias adjusted kappa; PPG, photoplethysmography; PPV, positive predictive value; PSG, polysomnography; REM, rapid eye movement; SD, standard deviation; SE, sleep efficiency; SOL, sleep onset latency; TIB, time in bed; TST, total sleep time; WASO, wake after sleep onset.

Disclaimer I am a military service member or employee of the US Government. This work was prepared as part of my official duties. Title 17, U.S.C. §105 provides that copyright protection under this title is not available for any work of the US Government. Title 17, U.S.C. §101 defines a US Government work as work prepared by a military service member or employee of the US Government as part of that person’s official duties. Report No. 21-62 was supported by the Office of Naval Research, Code 34, under work unit no. N1701. The views expressed in this article reflect the results of research conducted by the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, nor the US Government. The study protocol was approved by the Naval Health Research Center Institutional Review Board in compliance with all applicable federal regulations governing the protection of human subjects. Research data were derived from an approved Naval Health Research Center Institutional Review Board protocol, number NHRC.2017.0008.

Disclosure Evan D Chinoy, Joseph A Cuellar, and Jason T Jameson are employees of Leidos, Inc. The authors report no potential conflicts of interest in this work. None of the authors have any relationships related to the companies whose devices were evaluated in the study, nor were any companies involved in any stage of the research.

References

1.IndustryARC. Wearable Sleep trackers market: industry analysis, market size, share, trends, growth and forecast 2021-2026; 2021. Available from: https://www.industryarc.com/Report/19669/wearable-sleep-trackers-market.html. Accessed July 1, 2021.

2.Depner CM, Cheng PC, Devine JK, et al. Wearable technologies for developing sleep and circadian biomarkers: a summary of workshop discussions. Sleep. 2020;43(2):847. doi: 10.1093/sleep/zsz254 [DOI] [PMC free article] [PubMed] [Google Scholar]

3.Khosla S, Deak MC, Gault D, et al. Consumer Sleep Technology: an American Academy of Sleep Medicine Position Statement. J Clin Sleep Med. 2018;14(5):877–880. doi: 10.5664/jcsm.7128 [DOI] [PMC free article] [PubMed] [Google Scholar]

4.Schutte-Rodin S, Deak M, Khosla S, et al. Evaluating consumer and clinical sleep technologies: an American Academy of Sleep Medicine update. J Clin Sleep Med. 2021. doi: 10.5664/jcsm.9580 [DOI] [PMC free article] [PubMed] [Google Scholar]

5.Altini M, Kinnunen H. The Promise of Sleep: a Multi-Sensor Approach for Accurate Sleep Stage Detection Using the Oura Ring. Sensors. 2021;21(13):4302. doi: 10.3390/s21134302 [DOI] [PMC free article] [PubMed] [Google Scholar]

6.Chee NI, Ghorbani S, Golkashani HA, Leong RL, Ong JL, Chee MW. Multi-Night Validation of a Sleep Tracking Ring in Adolescents Compared with a Research Actigraph and Polysomnography. Nat Sci Sleep. 2021;13:177–190. doi: 10.2147/NSS.S286070 [DOI] [PMC free article] [PubMed] [Google Scholar]

7.Chinoy ED, Cuellar JA, Huwa KE, et al. Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep. 2021;44(5):zsaa291. doi: 10.1093/sleep/zsaa291 [DOI] [PMC free article] [PubMed] [Google Scholar]

8.de Zambotti M, Rosas L, Colrain IM, Baker FC. The Sleep of the Ring: comparison of the ŌURA Sleep Tracker Against Polysomnography. Behav Sleep Med. 2017;1–15. doi: 10.1080/15402002.2017.1300587 [DOI] [PMC free article] [PubMed] [Google Scholar]

9.de Zambotti M, Goldstone A, Claudatos S, Colrain IM, Baker FC. A validation study of Fitbit Charge 2^TM compared with polysomnography in adults. Chronobiol Int. 2018;35(4):465–476. doi: 10.1080/07420528.2017.1413578 [DOI] [PubMed] [Google Scholar]

10.Kahawage P, Jumabhoy R, Hamill K, de Zambotti M, Drummond SPA. Validity, potential clinical utility, and comparison of consumer and research-grade activity trackers in Insomnia Disorder I: in-lab validation against polysomnography. J Sleep Res. 2020;29(1):e12931. doi: 10.1111/jsr.12931 [DOI] [PubMed] [Google Scholar]

11.Lee XK, Chee NIYN, Ong JL, et al. Validation of a Consumer Sleep Wearable Device With Actigraphy and Polysomnography in Adolescents Across Sleep Opportunity Manipulations. J Clin Sleep Med. 2019;15(9):1337–1346. doi: 10.5664/jcsm.7932 [DOI] [PMC free article] [PubMed] [Google Scholar]

12.Menghini L, Yuksel D, Goldstone A, Baker FC, de Zambotti M. Performance of Fitbit Charge 3 against polysomnography in measuring sleep in adolescent boys and girls. Chronobiol Int. 2021;38(7):1010–1022. doi: 10.1080/07420528.2021.1903481 [DOI] [PMC free article] [PubMed] [Google Scholar]

13.Miller DJ, Roach GD, Lastella M, et al. A Validation Study of a Commercial Wearable Device to Automatically Detect and Estimate Sleep. Biosensors. 2021;11(6):185. doi: 10.3390/bios11060185 [DOI] [PMC free article] [PubMed] [Google Scholar]

14.Stone JD, Rentz LE, Forsey J, et al. Evaluations of Commercial Sleep Technologies for Objective Monitoring During Routine Sleeping Conditions. Nat Sci Sleep. 2020;12:821–842. doi: 10.2147/NSS.S270705 [DOI] [PMC free article] [PubMed] [Google Scholar]

15.de Zambotti M, Cellini N, Goldstone A, Colrain IM, Baker FC. Wearable Sleep Technology in Clinical and Research Settings. Med Sci Sports Exerc. 2019;51(7):1538–1557. doi: 10.1249/MSS.0000000000001947 [DOI] [PMC free article] [PubMed] [Google Scholar]

16.Haghayegh S, Khoshnevis S, Smolensky MH, Diller KR, Castriotta RJ. Accuracy of Wristband Fitbit Models in Assessing Sleep: systematic Review and Meta-Analysis. J Med Internet Res. 2019;21(11):e16273. doi: 10.2196/16273 [DOI] [PMC free article] [PubMed] [Google Scholar]

17.Scott H, Lack L, Lovato N. A systematic review of the accuracy of sleep wearable devices for estimating sleep onset. Sleep Med Rev. 2020;49:101227. doi: 10.1016/j.smrv.2019.101227 [DOI] [PubMed] [Google Scholar]

18.Lujan MR, Perez-Pozuelo I, Grandner MA. Past, Present, and Future of Multisensory Wearable Technology to Monitor Sleep and Circadian Rhythms. Front Digit Health. 2021;3:721919. doi: 10.3389/fdgth.2021.721919 [DOI] [PMC free article] [PubMed] [Google Scholar]

19.Lee J, Finkelstein J. Consumer sleep tracking devices: a critical review. Stud Health Technol Inform. 2015;210:458–460. [PubMed] [Google Scholar]

20.Markwald RR, Bessman SC, Reini SA, Drummond SPA. Performance of a Portable Sleep Monitoring Device in Individuals with High Versus Low Sleep Efficiency. J Clin Sleep Med. 2016;12(1):95–103. doi: 10.5664/jcsm.5404 [DOI] [PMC free article] [PubMed] [Google Scholar]

21.Meltzer LJ, Hiruma LS, Avis K, Montgomery-Downs H, Valentin J. Comparison of a Commercial Accelerometer with Polysomnography and Actigraphy in Children and Adolescents. Sleep. 2015;38(8):1323–1330. doi: 10.5665/sleep.4918 [DOI] [PMC free article] [PubMed] [Google Scholar]

22.Montgomery-Downs HE, Insana SP, Bond JA. Movement toward a novel activity monitoring device. Sleep Breath. 2012;16(3):913–917. doi: 10.1007/s11325-011-0585-y [DOI] [PubMed] [Google Scholar]

23.de Zambotti M, Cellini N, Menghini L, Sarlo M, Baker FC. Sensors Capabilities, Performance, and Use of Consumer Sleep Technology. Sleep Med Clin. 2020;15(1):1–30. doi: 10.1016/j.jsmc.2019.11.003 [DOI] [PMC free article] [PubMed] [Google Scholar]

24.Goldstein C. Current and Future Roles of Consumer Sleep Technologies in Sleep Medicine. Sleep Med Clin. 2020;15(3):391–408. doi: 10.1016/j.jsmc.2020.05.001 [DOI] [PubMed] [Google Scholar]

25.Rentz LE, Ulman HK, Galster SM. Deconstructing Commercial Wearable Technology: contributions toward Accurate and Free-Living Monitoring of Sleep. Sensors. 2021;21(15):5071. doi: 10.3390/s21155071 [DOI] [PMC free article] [PubMed] [Google Scholar]

26.Goldstein CA, Depner C. Miles to go before we sleep … a step toward transparent evaluation of consumer sleep tracking devices. Sleep. 2021;44(2):zsab020. doi: 10.1093/sleep/zsab020 [DOI] [PubMed] [Google Scholar]

27.Grandner MA, Rosenberger ME. Chapter 12 - Actigraphic sleep tracking and wearables: historical context, scientific applications and guidelines, limitations, and considerations for commercial sleep devices. In: Grandner MA editor. Sleep and Health. Academic Press; 2019:147–157. doi: 10.1016/B978-0-12-815373-4.00012-5. [DOI] [Google Scholar]

28.Menghini L, Cellini N, Goldstone A, Baker FC, de Zambotti M. A standardized framework for testing the performance of sleep-tracking technology: step-by-step guidelines and open-source code. Sleep. 2021;44(2):zsaa170. doi: 10.1093/sleep/zsaa170 [DOI] [PMC free article] [PubMed] [Google Scholar]

29.Carney CE, Buysse DJ, Ancoli-Israel S, et al. The Consensus Sleep Diary: standardizing Prospective Sleep Self-Monitoring. Sleep. 2012;35(2):287–302. doi: 10.5665/sleep.1642 [DOI] [PMC free article] [PubMed] [Google Scholar]

30.Arnal PJ, Thorey V, Debellemaniere E, et al. The Dreem Headband compared to polysomnography for electroencephalographic signal acquisition and sleep staging. Sleep. 2020;43(11):zsaa097. doi: 10.1093/sleep/zsaa097 [DOI] [PMC free article] [PubMed] [Google Scholar]

31.Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307–310. [PubMed] [Google Scholar]

32.Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8(2):135–160. doi: 10.1177/096228029900800204 [DOI] [PubMed] [Google Scholar]

33.Marino M, Li Y, Rueschman MN, et al. Measuring sleep: accuracy, sensitivity, and specificity of wrist actigraphy compared to polysomnography. Sleep. 2013;36(11):1747–1755. doi: 10.5665/sleep.3142 [DOI] [PMC free article] [PubMed] [Google Scholar]

34.Ancoli-Israel S, Cole R, Alessi C, Chambers M, Moorcroft W, Pollak CP. The role of actigraphy in the study of sleep and circadian rhythms. Sleep. 2003;26(3):342–392. doi: 10.1093/sleep/26.3.342 [DOI] [PubMed] [Google Scholar]

35.Hamill K, Jumabhoy R, Kahawage P, de Zambotti M, Walters EM, Drummond SPA. Validity, potential clinical utility and comparison of a consumer activity tracker and a research-grade activity tracker in insomnia disorder II: outside the laboratory. J Sleep Res. 2020;29(1):e12944. doi: 10.1111/jsr.12944 [DOI] [PubMed] [Google Scholar]

36.Roberts DM, Schade MM, Mathew GM, Gartenberg D, Buxton OM. Detecting sleep using heart rate and motion data from multisensor consumer-grade wearables, relative to wrist actigraphy and polysomnography. Sleep. 2020;43(7):zsaa045. doi: 10.1093/sleep/zsaa045 [DOI] [PMC free article] [PubMed] [Google Scholar]

37.Kubala AG, Barone Gibbs B, Buysse DJ, Patel SR, Hall MH, Kline CE. Field-based Measurement of Sleep: agreement between Six Commercial Activity Monitors and a Validated Accelerometer. Behav Sleep Med. 2020;18(5):637–652. doi: 10.1080/15402002.2019.1651316 [DOI] [PMC free article] [PubMed] [Google Scholar]

38.Berry RB, Brooks R, Gamaldo CE, et al. The AASM Manual for the Scoring of Sleep and Associated Events: rules, Terminology, and Technical Specifications. Version 2.2. Am Acad Sleep Med. 2015;1:34. [Google Scholar]

39.Colvonen PJ, DeYoung PN, Bosompra NOA, Owens RL. Limiting racial disparities and bias for wearable devices in health science research. Sleep. 2020;43(10):987. doi: 10.1093/sleep/zsaa159 [DOI] [PMC free article] [PubMed] [Google Scholar]

 

Draft - 12/27/2024 3:57 AM

 

{EEG & Sleep Analysis Diagnostics}