A sleep tracker can tell you that you slept for eight hours, but that number leaves out a lot. Two nights with the same duration can feel completely different, and part of the reason is how that time was divided between light, deep and REM sleep.
The Ultrahuman Ring estimates those stages by watching what the body is doing throughout the night. It does not measure brain activity directly. Instead, it looks at three signals — pulse, movement and skin temperature — and uses the pattern they create together to estimate which stage of sleep you are most likely in.
That distinction is useful to keep in mind when reading the numbers in the Ultrahuman app.
Three signals, one sleep-stage estimate
The Ring’s sleep tracking starts with its sensors.
Pulse is captured through an optical sensor that detects changes in blood volume at the finger with every heartbeat. From that signal, the Ring can determine heart rate and beat-to-beat timing, which is used to derive heart rate variability, or HRV.
Movement comes from an accelerometer that records how much the finger moves and in which direction. Physical stillness is common across sleep stages, so movement provides an important part of the classification.
Skin temperature is tracked at the finger. During sleep onset, skin temperature generally rises as core body temperature falls, and its overnight pattern also provides information related to circadian timing.
The Ring combines these streams rather than relying on one signal in isolation. A machine-learning model analyses successive periods of the night and assigns them one of four labels: wake, light sleep, deep sleep or REM sleep.
What separates deep, REM and light sleep?
The different stages leave somewhat different physiological signatures, although they are not perfectly distinct.
During deep sleep, heart rate tends to be at its lowest point of the night. Movement is minimal, while the HRV pattern shows stronger parasympathetic activity.
REM sleep is also associated with relatively little physical movement, but the cardiovascular pattern is different. Heart rate timing becomes more irregular and brief movements can occur.
Light sleep sits between these patterns and generally accounts for the largest portion of a typical night’s sleep.
Then there is wake. Movement is an important clue here. Someone who is lying completely still while awake can look surprisingly similar to someone in light sleep when pulse data is considered on its own. Movement helps the model distinguish the two.
This is also why identifying REM and deep sleep is not simply a matter of looking at one measurement. The model is looking at how several signals change together.
What you actually see in the Ultrahuman app
The underlying sensor data is turned into a simpler view in the Ultrahuman app’s Sleep dashboard.
You can see your total sleep duration alongside the amount of time estimated to have been spent in each sleep phase. The dashboard also presents a restorative sleep share, which Ultrahuman uses to describe the combined proportion of deep and REM sleep.
Internal Ultrahuman data from more than 275,000 Ring members gives some context for how that share varies across age groups. Among members aged 18–29, restorative sleep accounts for roughly 40% of the night, compared with about 34% among those aged 60 and over. Deep sleep falls from 17.0% to 14.2% across those groups, while REM falls from 23.3% to 19.5%.
Those figures describe population-level patterns rather than targets that every individual needs to hit.
Why one night’s result doesn’t tell the whole story
Sleep architecture naturally varies from night to night. The Ring’s estimates also have limitations, particularly when the signals become harder to interpret.
A fragmented night, illness, alcohol, an irregular sleep schedule, naps or very short sleep can all make the patterns less clear. There can also be brief awakenings that you never remember, while the time it takes to actually fall asleep may feel shorter than it was.
That is why sleep-stage data makes more sense when viewed across several weeks.
Your own baseline is generally more useful for understanding changes than comparing one night’s numbers with a population average. If your restorative sleep share shifts after changing your bedtime, reducing late-night alcohol or keeping a more consistent wake time, the pattern becomes more meaningful when it persists over multiple nights.
What if the data doesn’t match how you feel?
It can happen.
You might wake up feeling refreshed despite seeing less deep or REM sleep than expected. Another night might show an apparently reasonable sleep breakdown while you feel exhausted.
Neither piece of information needs to be dismissed. The sensor estimate provides one view of the night, while your own experience provides another.
The useful question is often whether the two tend to move together over time, rather than whether a single night’s numbers look ideal.
Does the Ring measure brain activity?
No.
The Ring measures pulse, movement and skin temperature and uses those signals to infer the most likely sleep stage. Direct measurement of brain activity requires electroencephalography, or EEG, which is one of the components used in clinical sleep studies.
That difference is important because the Ring’s sleep-stage data should be treated as an estimate for wellness tracking rather than a direct measurement of brain activity.
How does it distinguish REM from deep sleep?
Both stages involve relatively little physical movement, so movement alone cannot tell them apart.
The cardiovascular signals provide more of the distinction. Deep sleep tends to have a lower, more stable heart rate and a stronger parasympathetic HRV pattern. During REM, beat-to-beat timing becomes more irregular and brief movements may appear.
Because the distinction is more difficult, Ultrahuman also reports deep and REM together as restorative sleep, alongside the individual stage percentages.
Do you have to tell the app when you go to bed?
No. The Ring automatically detects sleep onset and offset from its sensor signals, so there is no need to manually start and stop a sleep session.
That allows the overnight record to be built from the same continuous stream of data rather than relying on someone remembering to activate a tracking mode.
How should you read your sleep-stage numbers?
The most useful approach is to treat them as a way of spotting patterns rather than grading individual nights.
Look at how your sleep changes over several weeks. Compare your current numbers with your own previous baseline. Use restorative sleep as a broad view, then look at the individual deep and REM percentages when you want more detail.
And keep the numbers alongside the rest of the picture: how rested you feel, whether your schedule has changed, and what was different about the days leading up to a particular night.
The Ultrahuman Ring is ultimately estimating sleep stages from signals produced by the body. It can give you a useful picture of how your sleep is changing, but it is not the same thing as a clinical sleep study. A polysomnography test records brain activity directly and is performed in a supervised clinical setting for diagnostic purposes.
For everyday tracking, the value lies less in finding a perfect percentage and more in noticing what your own sleep tends to look like over time.
Also Read: Mini Tadka Pans for Everyday Indian Cooking from Amazon
