How does sleep staging work?
A polysomnography (PSG) study records several signals through the night. For staging, the most important are the electroencephalogram (EEG), the electrooculogram (EOG) and the submental electromyogram (EMG). The recording is cut into consecutive 30-second windows called epochs, and a scorer assigns exactly one stage to each epoch by reading those signals together.
The result is a stage for every epoch of the night. Plotted over time it becomes a hypnogram, and summary measures such as total sleep time are calculated from it.
Why are sleep epochs 30 seconds long?
Thirty-second epochs are the standard defined in the AASM scoring manual, and most sleep centres still score at that resolution. As the authors of the YASA staging algorithm note, the reference datasets used to train and test scoring software are scored in 30-second epochs too, so software works at the same resolution. (Vallat & Walker, 2021 (opens in a new tab))
What are the five sleep stages?
| Stage | What it is | Features reviewers look for |
|---|---|---|
| Wake (W) | Awake, including brief awakenings during the night | Alpha rhythm with eyes closed, eye blinks and movements, higher muscle tone |
| N1 | Light, transitional sleep at sleep onset | Low-amplitude, mixed-frequency EEG; slow rolling eye movements |
| N2 | Stable light sleep; usually the largest share of the night | Sleep spindles and K-complexes |
| N3 | Deep, slow-wave sleep | High-amplitude slow (delta) waves |
| REM (R) | Rapid-eye-movement sleep, associated with dreaming | Rapid eye movements, very low chin-muscle tone, low-amplitude mixed EEG |
Older scoring systems split deep sleep into stages 3 and 4; the current AASM scheme combines them as N3. The exact scoring criteria are set out in the AASM Scoring Manual (opens in a new tab), which sleep facilities use as their reference.
What is a hypnogram?
A hypnogram is a step chart of the stage assigned to each epoch across the night. It shows sleep architecture at a glance: how quickly sleep began, how often it was interrupted, when deep sleep occurred and when REM periods appeared.
- Total sleep time (TST): time spent in N1, N2, N3 and REM.
- Sleep efficiency: total sleep time as a share of time in bed (needs lights-off and lights-on times).
- Sleep onset latency: time from lights off to the first epoch of sleep.
- Wake after sleep onset (WASO): wake time between the first and last sleep epochs.
- REM latency: time from sleep onset to the first REM epoch.
Is sleep staging the same as scoring a sleep study?
No. Staging is one part of polysomnography scoring. The AASM manual also covers arousals, respiratory events, movements during sleep and cardiac events. A staging tool alone does not score apnea, compute an apnea-hypopnea index, or diagnose a sleep disorder. (AASM (opens in a new tab))
Who does sleep staging, and how long does it take?
Staging is usually done by registered sleep technologists and reviewed by physicians who interpret the study. It is careful work: the authors of the YASA algorithm report that visually scoring a single night can take a well-trained scorer up to two hours, and that N1 shows very low agreement even between human scorers. (Vallat & Walker, 2021 (opens in a new tab))
Software can propose a stage for every epoch so that people review rather than score from scratch. See AI-assisted sleep scoring for how that works and where human review still matters.
Frequently asked questions
How long is a sleep epoch?
A sleep epoch is 30 seconds long. Each 30-second epoch of the recording receives exactly one stage.
How many sleep stages are there?
Current AASM scoring uses five stages: Wake, N1, N2, N3 and REM. Older systems split deep sleep into stages 3 and 4.
What signals are used to stage sleep?
Sleep is staged mainly from the EEG (brain activity), the EOG (eye movements) and the chin EMG (muscle tone), recorded during polysomnography.