Ten topics
One review per topic, from caffeine half-lives to where a sleep app’s job ends. Start anywhere.
- Alcohol and sleepAlcohol is sedating, but its most consistent effect on sleep is to delay and reduce REM sleep, starting at about two drinks and growing with the dose.24 findings · 39 sources · 9 myths
- Pre-sleep arousal and wind-down techniquesBeing mentally or physically keyed up at bedtime (hyperarousal) is central to every major model of insomnia, and racing thoughts and worry are the most common reason people with insomnia give for lying awake.25 findings · 44 sources · 8 myths
- Behaviour change: what improves sleep and how to make it stickCBT-I is the first-line treatment for chronic insomnia in every major guideline, and fully automated digital versions work in large trials.25 findings · 46 sources · 9 myths
- Caffeine and sleepCaffeine keeps you alert by blocking adenosine, the signal of time spent awake that builds sleep pressure.24 findings · 46 sources · 10 myths
- Bedroom environmentThe bedroom affects sleep mainly through heat, noise and light, and to a lesser degree through partners, pets, phones, air quality and the mattress.25 findings · 48 sources · 7 myths
- Light, the circadian clock and sleep timingSleep timing comes from sleep pressure plus a body clock in the brain that light keeps in time.25 findings · 59 sources · 9 myths
- Measuring sleep, and where an app's job endsDusk measures sleep with a morning self-report, which matches the expert-consensus sleep diary that clinics and insomnia trials use.24 findings · 52 sources · 9 myths
- Sleep pressure, naps, exercise and daytime behaviourSleep pressure (the homeostatic drive) builds the longer you are awake and drains during sleep, and together with the body clock it sets when you can fall asleep.24 findings · 47 sources · 8 myths
- Sleep regularity and schedule consistencySleep regularity is how consistent your sleep and wake times are from day to day, separate from how long you sleep.23 findings · 41 sources · 6 myths
- Sleep supplementsMost sleep supplements have small effects or no reliable effect, and sleep guidelines recommend cognitive behavioural therapy for insomnia (CBT-I) over any pill.25 findings · 62 sources · 11 myths
What the evidence levels mean
Every finding gets one of five ratings, and the rating decides how Dusk is allowed to say it. Across the whole base, 162 of 244 findings are strong or moderate. The rest are said with a hedge, or not said as advice at all.
- Strong evidence
Consistent results from guidelines, meta-analyses or several well-run trials.
In Dusk Stated plainly.
- Moderate evidence
Good studies mostly agree, but there are fewer of them, or they are smaller or in narrower groups of people.
In Dusk Stated plainly, with its limits.
- Limited evidence
A few small studies, lab work or observational data, without trials that settle it.
In Dusk Hedged: “may”, “small studies suggest”.
- Mixed evidence
Studies of reasonable quality point in different directions.
In Dusk Hedged, and never used to promise a result.
- Insufficient evidence
Too little good evidence to say either way.
In Dusk Never presented as a benefit.
How Dusk uses the evidence
Everything traces to a finding. The reasons on protocol steps, the cutoffs and temperatures, and the explanations Dusk’s AI writes are each tied to a finding here, and each finding lists the papers behind it.
Wording follows the rating. Strong and moderate findings are said plainly. Limited or mixed ones get “may” or “small studies suggest”. Insufficient evidence is never offered as a benefit.
The app says it simply; the numbers live here. Each finding has a one-line version safe to show anyone, and underneath it the doses, sample sizes and effect sizes, and who it doesn’t apply to.
Some signs mean a doctor, not a routine. Loud snoring with pauses in breathing, months of poor sleep, or low mood alongside it are reasons to see a clinician. Dusk’s rule for signs like these is to point you there, not to adjust your protocol. See the full list.
Wellness, not treatment
Dusk is a sleep-habit and tracking tool for general wellness; it does not diagnose or treat any condition.
Most of Dusk’s evening protocol is sleep hygiene: light, temperature, caffeine and alcohol timing, a steady wake time, a calmer last hour. These habits help many people sleep better. They are not how long-lasting insomnia is treated, and Dusk doesn’t claim to treat it.
- Strong evidence
Healthy sleep habits help many people, but on their own they are not a treatment for long-term insomnia.
- Strong evidence
For long-lasting insomnia, the recommended first step is a structured talking therapy called CBT-I, which is available in person or online.
- Strong evidence
Doctors use the word insomnia for trouble sleeping on 3 or more nights a week for 3 months or more, with daytime effects — Dusk can't diagnose it, but a doctor can.
Measuring sleep, and where an app’s job ends and what actually changes sleep go into the detail.
86 popular claims, checked
All myths →How this was put together
Each topic was reviewed on its own, preferring clinical guidelines and consensus statements first, then meta-analyses and systematic reviews, then randomised trials, large cohorts, and lab studies last. Where a popular claim rests only on lab or mechanistic work, the review says so.
Every DOI was checked against Crossref, and 406 of 407 matched; the other resolves through a different DOI registry. The 27 sources without a DOI are guidelines and agency pages, checked by their address. A few numbers come from abstracts or secondary summaries where the full text wasn’t available; the reviews flag them.
The evidence base is versioned, and these pages are rebuilt from it, so what you read here is what the app and its AI are working from.