Should I Track My Sleep?

There is no shortage of technology that will help you track your sleep. Devices range from wearables to devices that sit on your nightstand or go under your mattress (nearables). The wrist- or finger-worn devices (wearables) at a minimum measure movement and often measure heart rate and skin temperature. Nearables remotely track physiological and behavioral signals (e.g., body movement, heart rate and breathing). Sleep tracking devices can be simply used to increase personal awareness of sleep and can be an important reminder to prioritize sleep.

There have been an increasing number of evaluations of the accuracy of these devices in recent years, many by the Naval Health Research Center. The gold standard of measuring sleep is polysomnography, where brain waves, muscle tone and eye movements are recorded. Registered Polysomnographic Technologists (RPSGTs) review each 30-second interval of brain wave amplitudes and frequencies and muscle movements and score it as awake, light sleep (Stage 1 and Stage 2), slow-wave, deep sleep (Stage 3) or Rapid Eye Movement (REM) sleep. It is important to note that the best RPSGTs agree ~85-90% of the time on the interval’s score.

Laboratory studies evaluating sleep tracking devices compare the automated estimated-sleep output of the tracking device with the PSG-measured sleep. These lab studies generally perform well because time in bed is controlled. The ability of the tracking device to recognize sleep versus awake is generally as good as the agreement between RPSGTs.

Recent, more challenging studies have been conducted in ‘real-world’ conditions. Given that it is difficult to conduct full PSG at home, the devices are compared against other PSG-like devices that estimate sleep (e.g., a band worn around your head). Therefore, the comparison is not as pure, but devices again perform fairly well in estimating sleep duration. Overall, the devices do not estimate wake as well as sleep. With wearables, it is difficult to distinguish between quiet wakefulness (e.g., lying in bed reading a book) and sleep. Wearables can also fail to detect short naps, and nearables do not recognize a nap unless it is in bed. Nearable data are also subject to misinterpretation when pets or partners share a bed. Further, although the tracking devices are pretty good at estimating your sleep duration, they are not yet as good at estimating the stages of sleep. Some studies show the determination of sleep stage with tracking devices is no better than chance.

All of these validation studies have been conducted on healthy subjects. Individuals with sleep disorders that fragment and disrupt sleep, such as obstructive sleep apnea or insomnia, are excluded. Further, studies have not been conducted with shift workers, who often nap and sleep during the biological day. It’s likely the devices are less accurate when encountering these more challenging work/sleep situations, such as railroad worker or airline pilot schedules.

In summary, the sleep tracking devices are not perfect, but have the potential to get better as machine learning algorithms become incorporated. Currently, the devices are good enough to increase your awareness of sleep, remind you to make sleep a priority and give you enough information to assess whether you are getting the recommended 7-9 hours of sleep each day. You can also observe your daily sleep and wake times to determine the consistency of your sleep. Although this can be very difficult for pilots, more studies are showing that sleep regularity is important to overall health.