Skip to content

Sleep not showing awake points #593

Open

Description

@markpiggott-ctrl

Which metric?

Sleep stages

What it showed, and why you think it's wrong

It doesn’t look like the sleep report is showing awake times. Last night (see image) I was awake at least 3 times for more than 15 minutes, but none are picked up. Whoop MG on iOS.

What kind of wrong?

Other

App version

0.10.0

Which band?

WHOOP 4.0

Screenshots

Image

Activity

  1. kurexw commented on Nov 13, 2026

    @kurexw
    Contributor

    Thanks for the screenshot. The chart itself looks right: the blank gaps between the Light blocks are short REM and Deep stretches. The problem is that staging only scored about 8 minutes awake that night (98% asleep of 5h 49m in bed).

    Sleep staging has changed several times since 0.10.0. On current main I put three 18-minute wake-ups into a synthetic WHOOP 4.0 night, and the awake time rose by about 40 minutes when the wrist was still and by about 80 when it was moving. So long wake-ups are being picked up now.

    Could you check a night with a clear wake-up on 0.11.3 or later and say whether it now shows as Awake? If it still misses one, the date and the rough times you were awake would help. Sending an in-app report (Settings → Report an issue) after that night would let us look at the actual data.

  2. abdulsaheel commented on Nov 17, 2026

    @abdulsaheel
    Collaborator

    sleep staging has changed a fair bit since 0.10.0, including how long wake-ups inside the night are handled. can you look at a night on 0.12.2 where you were clearly awake for 15+ minutes and tell me if it shows as awake now? if it still misses it, the date and rough times plus an in-app report would help.

  3. BucciMobile commented on Nov 18, 2026

    @BucciMobile
    Contributor

    Thanks for the report - we looked into this from the physiology side while working on the opt-in HMM stager (PR #756), and what you're seeing is a known weak spot of the default stager.
    Right now an epoch only counts as wake if movement AND elevated heart activity show up together. If you're lying in bed awake but still - say, scrolling on your phone - you barely move and your heart rate is only slightly up, so neither condition fires and the epoch gets scored as light sleep. That matches your screenshot.
    The tricky part is REM sleep: during REM the heart rate can actually be as high as or higher than when awake, and heart rate variability looks wake-like too (sympathetic dominance). So a simple "heart rate high = awake" rule would wrongly call REM awake. At the same time, "lying still and awake" is barely distinguishable from sleep on heart signals alone - just a few bpm difference, HRV hardly changes. Without EEG (which is what AASM scoring in a sleep lab uses) this is genuinely a gray zone.
    What the research says: for wrist devices, movement is by far the most reliable sleep/wake signal, heart rate on its own is weak for sleep vs wake (npj Digital Medicine 2024 review of 62 wearable setups). That's why our experimental stager in PR #756 treats movement as its own wake cue and only uses heart rate as a small, REM-guarded extra hint.
    For the default stager we'd suggest: wake = moving OR (still AND heart rate clearly above your personal sleep baseline), with REM protected by its own rules. Once we have reference nights to validate against PSG (kappa), we can port that into the analytics package.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions