Heart rate range demonstrates distinct variability patterns between quiet sleep and active sleep in preterm infants, with an area under the curve of 0.88 at a threshold of 18 bpm distinguishing the two states.
Key Findings
Results
ROC analysis effectively distinguished heart rate ranges between active sleep and quiet sleep in preterm infants with an AUC of 0.88.
The threshold for distinguishing active sleep from quiet sleep was 18 bpm.
Heart rate range during active sleep had a median of 27.597 bpm (minimum: 12.385, maximum: 106.705 bpm).
Heart rate range during quiet sleep had a median of 13.943 bpm (minimum: 5.261, maximum: 36.284 bpm).
HR variability was analyzed using a 30-second window with a 2-second sliding interval.
Reliability was confirmed by dividing the cohort into two groups and repeating the analysis, producing consistent findings.
Results
Preterm infants spent proportionally more time in active sleep than quiet sleep during recordings.
On average, infants spent 62% of recorded time in active sleep.
On average, infants spent 38% of recorded time in quiet sleep.
The study included 68 infants born between 23 and 36 weeks of gestational age.
EEG recordings were obtained between 34.857 and 40.714 weeks postmenstrual age (PMA).
Results
EEG delta power was significantly higher during quiet sleep compared to active sleep, independent of postmenstrual age and sex.
Delta frequency band analyzed was 0.5–4 Hz.
The difference in delta power between quiet sleep and active sleep was statistically significant (P < 0.05).
Linear mixed-effects models were used to examine the dependence of EEG metrics on sleep state.
Sex and postmenstrual age at study were included as covariates in the model.
Higher delta power during quiet sleep is consistent with known cortical correlates of this sleep state.
Methods
The study cohort consisted of preterm infants spanning a wide gestational age range with recordings performed at near-term and term-equivalent ages.
Infants were born between 23 and 36 weeks of gestational age (n = 68).
EEG and ECG recordings were obtained between 34.857 and 40.714 weeks PMA.
Video EEG and ECG data were collected concurrently, allowing visual behavioral state classification to be paired with physiological measures.
Videos were reviewed visually for active sleep and quiet sleep states by human observers.
Discussion
Validation of heart rate range findings in younger preterm infants is currently underway.
The current cohort included recordings from 34.857 weeks PMA and above.
The authors noted that extension to younger preterm infants is needed to generalize findings.
This limitation was acknowledged as a direction for future work.
What This Means
This research suggests that the variation in a premature baby's heart rate can reliably indicate which sleep state the baby is in — either active sleep (similar to REM sleep) or quiet sleep (similar to deep sleep). By analyzing how much a baby's heart rate fluctuates within short 30-second windows, researchers found that heart rate swings more widely during active sleep (average range of about 28 beats per minute) compared to quiet sleep (average range of about 14 beats per minute). A cutoff of 18 beats per minute distinguished the two states with high accuracy, achieving an area under the curve of 0.88, where 1.0 would be perfect classification.
The study also found that brain wave activity, specifically slow 'delta' waves measured by EEG, was stronger during quiet sleep than active sleep, which is consistent with what is known about sleep biology. Together, the heart rate and brain wave findings reinforce each other and suggest that simple heart rate monitoring could serve as a practical tool for tracking sleep states in premature infants, who are too fragile for more invasive assessments.
This matters because sleep is critical for brain development in premature babies, and being able to accurately identify sleep states without complex equipment could help clinicians better monitor and support these vulnerable infants. The researchers studied 68 premature babies born as early as 23 weeks of pregnancy and confirmed their findings by testing them in split groups. They note that further work is needed to see if these findings hold in even younger premature infants.
Govindan R, Chirumamilla V, Mulkey S, Baker R, du Plessis A. (2026). Identifying heart rate characteristics of sleep states of preterm infants using video analysis.. Scientific reports. https://doi.org/10.1038/s41598-025-31873-7