Sleep Regularity as a Digital Biomarker

The Sleep Regularity Index quantifies how consistently a person's sleep and wake timing repeats from one day to the next. It is calculated as the probability that a person is in the same state, asleep or awake, at the same clock time on two consecutive days, scaled so that perfectly regular sleep scores 100 and completely random sleep scores zero.

Its importance is that it is orthogonal to duration. Two people can average identical total sleep time while one sleeps the same hours every night and the other alternates between early and late, and every measure in this library that reports a nightly average treats them as equivalent.

The strongest finding in this area is that regularity predicts mortality more strongly than duration does, which inverts the emphasis of most sleep advice and of most sleep tracking products.

Calculation requires continuous sleep and wake classification across many consecutive days, which is why the measure only became practical with wearables. Actigraphy or wearable sleep staging produces a binary sleep or wake state at fine time resolution, and the index compares each epoch against the same epoch twenty four hours earlier across the whole recording.

A minimum of about a week is needed and more is better, because the statistic is about consistency and a short window cannot distinguish a genuinely regular sleeper from someone observed during a stable stretch. Open source implementations exist and have been applied at scale, including to tens of thousands of accelerometer records in a national biobank.

Several competing regularity metrics exist, and their theoretical properties and practical behaviour differ enough that the choice matters and should be reported.

The main research use is as an exposure in epidemiological work linking sleep behaviour to cardiometabolic, psychiatric and mortality outcomes, where it has repeatedly outperformed duration. It is also used as an outcome in circadian and behavioural sleep interventions, and in shift work research where irregularity is the exposure of interest.

In mental health research regularity is attractive because it captures behavioural disorganisation without requiring the person to report anything, and irregular sleep and wake patterns have been associated with mood and functional outcomes.

It is reported alongside the Pittsburgh Sleep Quality Index and the Insomnia Severity Index, which capture perceived sleep quality. The pairing is informative precisely because a person can be highly irregular and report sleeping well.

The measure inherits every weakness of the underlying sleep and wake classification, and consumer devices are least accurate at detecting wakefulness, which is exactly what the index depends on.

It is also indifferent to what regularity is achieved. Consistently sleeping at biologically inappropriate times scores as regular, so a night shift worker with a fixed schedule can score highly while carrying substantial circadian risk. Regularity and circadian alignment are different things and this index measures only the first.

Missing data is a practical problem, since the calculation compares matched epochs across days and gaps propagate. Finally, the metric is newer than most in this library, so while the epidemiology is strong, there is no established threshold defining what counts as clinically meaningful irregularity.

• Windred DP, et al. Sleep regularity is a stronger predictor of mortality risk than sleep duration: a prospective cohort study. Sleep. 2024. pubmed.ncbi.nlm.nih.gov
• Phillips AJK, et al. Irregular sleep/wake patterns are associated with poorer academic performance and delayed circadian and sleep/wake timing. Sci Rep. 2017. pubmed.ncbi.nlm.nih.gov
• Windred DP, et al. Objective assessment of sleep regularity in 60,000 UK Biobank participants using an open-source package. Sleep. 2021. pubmed.ncbi.nlm.nih.gov
• Fischer D, et al. Measuring sleep regularity: theoretical properties and practical usage of existing metrics. Sleep. 2021. pubmed.ncbi.nlm.nih.gov