Sleep Onset Latency as a Digital Biomarker

Sleep onset latency is the time between attempting to sleep and actually falling asleep. It is the measure that isolates sleep initiation, as distinct from sleep maintenance, and it is the defining complaint of a large group of people with insomnia who fall asleep with difficulty but then sleep through. Latency beyond roughly thirty minutes is the threshold most commonly used to define clinically significant difficulty initiating sleep, and it appears in the diagnostic criteria and outcome definitions used across insomnia research. Sleep onset latency is also one of the three core outcomes recommended by consensus for insomnia trials, alongside wake after sleep onset and sleep efficiency. As a digital biomarker it is estimated every night at home rather than recalled at a clinic visit, which matters because the recall of how long it took to fall asleep is unusually distorted in exactly the people who complain about it.

Actigraphy identifies sleep onset as the first sustained run of inactivity after the start of the sleep period, and latency is the interval between that point and the start point. Devices with photoplethysmography add the fall in heart rate and change in heart rate variability that accompany sleep onset, which sharpens the estimate. The critical dependency is the start point itself. Polysomnography, the reference standard, has an unambiguous lights out marker set by a technician. A wearable has to infer when the participant intended to sleep, either from a sleep diary, from a user-entered bedtime, or from an algorithm reading stillness and context. Any error in that inference goes directly into the latency estimate, which is why sleep onset latency is the standard sleep measure most sensitive to how the sleep period is defined.

Sleep onset latency is a standard endpoint in insomnia trials, particularly for drugs and behavioural therapies targeting sleep initiation, and it is a monitoring measure in anxiety, chronic pain and paediatric research where difficulty settling is a presenting problem. It is also used in circadian and shift work studies, where a lengthening latency is one of the earliest signs that sleep timing has drifted out of alignment with the body clock. In the public DiMe endpoint library it is registered as a secondary endpoint in sleep and menopause trials assessed by actigraphy. Because the objective and subjective versions of this measure diverge so markedly, trials almost always collect the Insomnia Severity Index alongside it, and interpret the wearable estimate as a complement to the questionnaire rather than a replacement for it.

Sleep onset latency depends entirely on knowing when the attempt to sleep began, and wearables estimate that boundary rather than observe it, so absolute values carry more uncertainty than any other standard sleep measure. Actigraphy also tends to score early quiet wakefulness as sleep, which shortens the estimate, and the shortening is largest in people with insomnia. Reading in bed, screen use and other quiet pre-sleep activity all confound the signal. Values should not be compared across device brands or across studies that define the sleep period differently, and change within a participant is far more trustworthy than the absolute number.

• Smith MT, et al. Use of actigraphy for the evaluation of sleep disorders and circadian rhythm sleep-wake disorders: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med. 2018. pubmed.ncbi.nlm.nih.gov
• Buysse DJ, et al. Recommendations for a standard research assessment of insomnia. Sleep. 2006. pubmed.ncbi.nlm.nih.gov
• Chinoy ED, et al. Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep. 2021. pubmed.ncbi.nlm.nih.gov
• Bastien CH, Vallieres A, Morin CM. Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Med. 2001. pubmed.ncbi.nlm.nih.gov