5 Tips for Selecting a Sleep Wearable

Quality sleep is critical for physical and mental health, and its benefits for athletic performance are well established. The strain of COVID-19 lockdowns has manifested in worse sleep behaviors and compromised well-being, even among top athletes. Given heightened health vigilance, increased time at home, and the added stress pandemic life, it makes sense that athletes of all levels are turning towards sleep wearables for health insights.

Monitoring sleep with wearables is not new, but the hardware and software used by these devices have improved noticeably in recent years. In turn, so have their ability to provide more and better metrics, paired with semi-customized advice on sleep behavior modification. It can be daunting to wade through the piles of marketing materials to figure out the best device. While I’m not a sleep expert, I’ve been fortunate to work with some of the top researchers in the field.

Here is wisdom I’ve gleaned from top sleep wearables experts to help me (and you) in the search for the best device. 

Some Basics

Wearables vs. Nearables

Sleep monitoring typically comes in two flavors. Wearables, which are worn on the body during sleep, and nearables, which are placed on or near the bed. Wearables are most commonly worn on the wrist but also appear in the form of armbands, rings, and headbands. Nearables include smart mattresses, mattress attachments, and bedside monitors. The tips below are geared towards wearables but also can apply to nearables.

How Sleep Wearables Work

Earlier generations of devices were built on the principles of actigraphy and relied primarily on movement detected by accelerometers to determine sleep vs. wake. In general the current devices on the market augment accelerometry with additional sensors, namely photoplethysmography (PPG) although skin temperature and electroencephalography (EEG) sensors are also present in some devices. The current generation of multi-sensor devices is expected to provide more accurate classification of sleep vs. wake and sleep staging, but the research is evolving.

What They Measure

Most wearables provide some or all of the “standard” sleep metrics, including:

• Total Sleep Time (TST): accumulated time asleep during the night 

• Wake After Sleep Onset (WASO): accumulated time awake after initially falling asleep

• Time in Bed (TIB): how long the individual was in bed, whether sleeping or not. Essentially “lights off” to “lights on”.

• Sleep Onset Latency (SOL): how long it took the individual to fall asleep after getting into bed

• Sleep Efficiency (SE): percentage of time in bed actually asleep, which is TST divided by TIB

• Sleep stages: usually categorized as REM, “Light” (combination of non-REM stage 1 and stage 2), and “Deep” (non-REM stage 3)

Non-Sleep Measures

Devices incorporating PPG typically provide information about heart rate, heart rate variability (HRV), and/or breathing rate during sleep, since these data can be derived from PPG. More recently some devices have started estimating oxygen saturation, and at least one device provides information on body temperature changes during sleep. I’ll save discussion of these measures for another time and focus mainly on sleep, acknowledging that some of these measures may be part of a device’s sleep-detection algorithms.

Expert Advice for Selecting a Sleep Wearable

Now that you have the basics, here are some key points to see in mind when comparing wearable devices.

1. All Wearables are not created equal

Examining the published research and talking to experts in the field reveals that there is a wide variability in how accurately wearable devices measure sleep. Since many devices are coming out with new versions of software or hardware more than once per year, the accuracy also can vary substantially within a device model. Bad data in means bad data out, so it’s worthwhile spending time to investigate the quality of any device’s data. Read this to learn more about evaluating device validity.

2. Wearables work best for “normal” sleepers

Most devices assume that an individual has their main bout of sleep during the night. As a result, wearables tend to be less accurate at detecting daytime naps or main bouts of sleep that occur during the daytime. Shift workers and individuals who cross multiple time zones likely will see less accurate data. Additionally, devices tend to be less accurate in individuals with sleep disorders, such as obstructive sleep apnea. 

3. Total Sleep is more accurate than Total Wake or Sleep Staging

In terms of accuracy, devices tend to be best with measuring Total Sleep Time although they generally still overestimate it. Since almost all of time in bed is sleep, particularly for good sleepers, even if a device has no detection algorithm at all and just calls everything sleep, it will get a good estimate of TST. 

Wearables tend to underestimate Total Wake / Wake After Sleep Onset, and this may be worse in individuals with disordered sleep. It is challenging to distinguish periods of quiet in-bed wakefulness from actual sleep is non-trivial. Most validation studies also do not evaluate Time in Bed or Sleep Onset Latency, which require reliable automatic detection of “lights off,” sleep onset, and “lights on.”

Wearables tend to be more robust in estimating time spent in “light” and REM sleep stages compared to “deep” sleep, but accuracy varies substantially across devices. As with Wake, the gold standard for staging sleep, polysomnography, relies on several types of data (e.g., EEG, EOG, EMG) not typically present in a sleep wearable. Sophisticated algorithms have been developed to infer sleep stage from accelerometry, PPG, and other data sources. While these algorithms continue to improve, they are generally not very accurate at this time.

4. Don’t put too much stock in Sleep Score 

Many devices on the market provide versions of a “Sleep Score” that numerically conveys the quality of the previous night’s sleep. Such a score does not exist in standard sleep research or clinical practice, and there is no consensus among sleep experts about how to rate the quality of one’s sleep with data. Any “Sleep Score” is a non-standard calculation made by the manufacturer using proprietary algorithms that may or may not be grounded in sleep science. Few if any of these “scores” have been validated in a research study to confirm that they relate to meaningful sleep or performance outcomes. 

5. There is no Swiss Army knife

Most current sleep wearables offer measurements beyond sleep, such as activity tracking or heart rate measurements. Each measurement requires its own combination of sensors and sophisticated algorithms. It is rare and likely impractical to find a device able to perform optimally across all categories. Thus, most wearables experts recommend leveraging a device specifically designed for sleep and leaving the tracking of other metrics to devices tailored to those measurements. 

Summary

Athletes, coaches, and weekend warriors are turning to wearables to improve their sleep and wellness. Sleep wearables vary widely in the quality of their data and, in turn, the value of the insights they provide. Take some time to investigate device accuracy. Be aware that even the best devices may struggle with correctly identifying wake and distinguishing sleep stages. If you are a shift worker, frequent flyer, or have a sleep-related disorder, a sleep wearable may not be your best bet at this time. Regardless, look for a dedicated sleep device rather than an “all-arounder” and plan to spend time learning your own baseline so that you do not waste energy fretting over arbitrary “scores.” Happy sleeping!

Extra Credit

Check out the articles below to learn more about sleep wearables, device accuracy, and monitoring sleep in athletic populations. I’ve also written a post about how to read and interpret sleep validation research.

C. M. Depner, P. C. Cheng, et al., “Wearable technologies for developing sleep and circadian biomarkers: A summary of workshop discussions,” Sleep, vol. 43, no. 2, Feb. 2020. (open access)

M. de Zambotti, N. Cellini, et al., “Wearable sleep technology in clinical and research settings.,” Med Sci Sports Exerc, vol. 51, no. 7, pp. 1538–1557, Jul. 2019. (open access)

E. R. Facer-Childs, D. Hoffman, et al., “Sleep and mental health in athletes during COVID-19 lockdown.,” Sleep, Feb. 2021. (open access)

C. Goldstein, “Current and future roles of consumer sleep technologies in sleep medicine.,” Sleep Med Clin, vol. 15, no. 3, pp. 391–408, Sep. 2020.

R. K. Malhotra, “Sleep, recovery, and performance in sports.,” Neurol Clin, vol. 35, no. 3, pp. 547–557, Aug. 2017.

H. Scott, L. Lack, and N. Lovato, “A systematic review of the accuracy of sleep wearable devices for estimating sleep onset,” Sleep Medicine Reviews, vol. 49, p. 101227, Feb. 2020.

J. D. Stone, L. E. Rentz, J. Forsey, J. Ramadan, R. R. Markwald, V. S. Finomore, S. M. Galster, A. Rezai, and J. A. Hagen, “Evaluations of commercial sleep technologies for objective monitoring during routine sleeping conditions.,” Nat Sci Sleep, vol. 12, pp. 821–842, 2020. (open access)

N. P. Walsh, S. L. Halson, et al. , “Sleep and the athlete: narrative review and 2021 expert consensus recommendations.,” Br J Sports Med, Nov. 2020. (open access)