The Owlet Dream Sock determines deep sleep by combining two sensors on your baby’s foot: a pulse oximeter that reads heart rate and blood oxygen through the skin, and an accelerometer that tracks movement. The device uses changes in these signals to estimate whether your baby is in deep sleep, light sleep, or active (REM) sleep at any given moment.
The Sensors on Your Baby’s Foot
The sock wraps around your baby’s foot and shines light through the skin to measure pulse rate and oxygen levels, a technique called photoplethysmography (PPG). This is the same basic technology used in hospital pulse oximeters and adult fitness trackers, scaled down for an infant’s foot. Alongside the light sensor sits a small accelerometer that detects motion. If the accelerometer picks up too much movement, the device pauses its readings rather than reporting inaccurate data.
These two data streams, heart rate patterns and physical stillness or movement, form the raw inputs the Owlet uses to sort your baby’s sleep into stages.
How Heart Rate and Movement Signal Sleep Stages
During deep sleep (called “quiet sleep” in infants), heart rate drops and becomes very steady, breathing slows, and the body is nearly motionless. During active sleep, which is the infant equivalent of REM, heart rate is more variable, breathing is irregular, and babies twitch, grimace, or move their limbs. Light sleep falls somewhere in between, with moderate heart rate variability and occasional small movements.
The Owlet’s algorithm looks for these patterns in real time. A sustained drop in heart rate combined with minimal accelerometer movement signals deep sleep. Rising heart rate variability paired with detected limb movement signals active sleep. The device assigns a sleep stage label based on whichever pattern best matches the incoming data, then displays it in the companion app as a color-coded timeline.
Owlet has not published the exact thresholds or weighting its algorithm uses, so the precise cutoff between “deep” and “light” in its system is proprietary. What’s known from the FDA review of the Dream Sock is that the device relies on the PPG sensor for pulse rate and oxygen data and the accelerometer for motion quality assessment.
Why Infant Sleep Stages Are Harder to Track
Infant sleep is structured very differently from adult sleep, which matters for how accurately any consumer device can classify it. Newborns have three sleep states: quiet sleep (similar to deep sleep), active sleep (similar to REM), and indeterminate sleep, a transitional state that doesn’t fit neatly into either category. Adults cycle through sleep stages in roughly 90-minute blocks, but infant sleep cycles last only about 50 minutes, and newborns enter sleep through REM rather than deep sleep.
For the first two to three months of life, babies don’t even have established circadian rhythms. Their sleep is distributed almost evenly across day and night, with the longest stretch typically lasting only 2.5 to 4 hours. Circadian patterns start emerging around two to three months, when melatonin and cortisol begin cycling predictably and sleep onset shifts from REM-first to deep-sleep-first. This means the Owlet is trying to classify sleep stages that are inherently messier and shorter than adult stages, and the patterns it’s looking for change as your baby matures.
Accuracy Compared to Clinical Sleep Studies
The gold standard for sleep staging is polysomnography, a clinical sleep study that uses brain wave monitoring (EEG), eye movement tracking, and muscle activity sensors. The Owlet doesn’t have any of these. It’s inferring sleep stages from peripheral signals (heart rate and limb movement), which correlate with sleep stages but don’t measure them directly.
Owlet has not published a peer-reviewed validation study comparing its sleep stage classifications against polysomnography. The clinical data that does exist for the device focused on its ability to detect low oxygen levels and abnormal heart rates, not sleep staging accuracy. In that context, the sock showed 88.8% sensitivity and 85.7% specificity for detecting low blood oxygen. Sleep stage accuracy is a separate question, and without published validation data, it’s not possible to assign a specific accuracy percentage to the deep sleep readings you see in the app.
The Dream Sock received FDA Class II clearance in November 2023 as an “infant pulse rate and oxygen saturation monitor for over-the-counter use.” That clearance covers the device’s ability to measure heart rate and oxygen, not its sleep staging feature. The sleep tracking is treated as a wellness feature, not a medical one.
What the Sleep Data Means in Practice
When the Owlet app shows your baby spent a certain number of minutes in deep sleep, it’s showing an estimate based on heart rate stability and physical stillness. It’s reasonable as a general guide to your baby’s sleep patterns over time. You can use it to spot trends: whether deep sleep stretches are getting longer as your baby ages, whether naps tend to be mostly light sleep, or whether nighttime wake-ups are disrupting deeper sleep phases.
Where it’s less reliable is in moment-to-moment precision. A brief arm twitch might cause the algorithm to reclassify a deep sleep period as light sleep, or a still but wakeful baby might register as sleeping. The shorter and more irregular your baby’s sleep cycles are (especially under three months), the harder it is for any motion-and-heart-rate-based system to draw clean lines between stages. As your baby’s circadian rhythms mature and sleep cycles lengthen, the device’s classifications generally become more consistent.