You
Today's schedule
Caffeine
Alcohol
Exercise
Tonight's sleep
Key numbers
Predicted hypnogram
Sleep architecture
Two-process model
Your 24-hour clock
Each ring shows when something happens. Inner glow = sleep pressure. Outer dots = events. Background gradient = circadian (your body's clock).
Legend
- Caffeine doses (size = mg)
- Alcohol drinks (size = drinks)
- Exercise bouts (length = duration)
- Last meal
- Bright light exposure
- Predicted sleep window
- Body temperature nadir (CBTmin)
- Predicted DLMO (melatonin onset)
What it tells you
- Caffeine within ~ 8 h of bed leaves residual; visible as a long arc.
- Alcohol in the evening shifts REM toward the second half.
- Bright light close to bed delays the sleep window.
- The grey-shaded arc shows where your body wants to be asleep (S above threshold), ideally aligned with your bedtime.
Tomorrow's alertness
SAFTE effectiveness curve from this morning's wake through tomorrow evening. Lower KSS = more alert.
Tomorrow's peak windows
What would improve tonight?
Each intervention re-runs the full forecast and shows how much your sleep score improves.
How SleepArchitect works
Sleep researchers have a 40-year-old quantitative model, Borbely's two-process model, that predicts sleep timing from just two variables: Process S (homeostatic sleep pressure, rising linearly during wake and falling exponentially during sleep) and Process C (circadian propensity, a sinusoid set by your body clock). Adding caffeine pharmacokinetics, alcohol pharmacokinetics, exercise effects, meal-timing effects, and light-PRC-based circadian phase shifts gives a complete personalised sleep predictor.
Existing sleep apps (Oura, Whoop, Apple Health, Sleep Cycle, AutoSleep) track sleep retrospectively. SleepArchitect is, to our knowledge, the first free open-source tool that predicts tonight's sleep from today's caffeine, alcohol, exercise, light, and meal timing.
Citations
- Borbely 1982 Hum Neurobiol: two-process model
- Daan, Beersma & Borbely 1984 Am J Physiol: tau parameters
- Skeldon et al. 2017 Sci Rep: two-process implementation
- Khalsa et al. 2003 J Physiol: light PRC
- Zeitzer et al. 2000 J Physiol: light dose-response
- Brainard 2001 J Neurosci, Lockley 2003: melatonin suppression
- Drake et al. 2013 JCSM: caffeine effect on sleep
- Clark & Landolt 2017 Sleep Med Rev: caffeine + SWS
- Ebrahim et al. 2013 ACER: alcohol + REM
- Stutz et al. 2019 Sleep Med Rev: exercise meta-analysis
- Crispim 2011, Iao 2021: meal timing
- Hursh / Mallis 2004: SAFTE alertness model
- Roenneberg 2003 J Biol Rhythms: MCTQ chronotype
- Akerstedt & Gillberg 1990 Int J Neurosci: KSS
Privacy
SleepArchitect runs entirely in your browser. No analytics, no
cookies, no network calls. Profile is stored only in your browser's
localStorage.
Limitations
This is a forecast not a sleep study. Sleep architecture predictions are population averages, not personalised polysomnography. For clinical sleep concerns (apnoea, insomnia, RLS, narcolepsy), see a sleep physician. Inputs assume non-pregnant healthy adults; use pregnancy and CYP1A2 toggles for relevant covariates.