Article Text

Download PDFPDF

P006 Modelling sleep-wake transitions in very and moderately pre-term infants
  1. Anastasis Georgoulas1,
  2. Asif Tamuri1,
  3. Maria Pureza Laudiano-Dray2,
  4. Judith Meek3,
  5. Lorenzo Fabrizi2 and
  6. Kimberley Whitehead2
  1. 1Research IT Services, University College London, London, UK
  2. 2Dept. of Neuroscience, Physiology and Pharmacology, University College London, London, UK
  3. 3Elizabeth Garrett Anderson Wing, University College London Hospitals, London, UK


Introduction Sleep is the dominant vigilance state in pre-term infants, but its regulation is still poorly understood, with no underpinning quantitative framework.

In animal studies, neonatal sleep-wake characteristics follow statistical patterns, e.g. wakefulness durations exhibit an exponential distribution.1 Here we investigated whether the same holds true for pre-term human infants.

Methods We recorded electroencephalography (EEG), respiratory movement, electrocardiography (ECG), and behavioural observations for up to two hours from 54 non-mechanically ventilated infants being cared for on the neonatal unit (28+2–34+1 weeks+days corrected gestational age). Data were staged as sleep or wakefulness in 30-s epochs (figure 1).

We characterised i) the distribution of wakefulness durations, using the Kolmogorov-Smirnov goodness-of-fit test, and ii) the likelihood of transitioning from sleep to wakefulness during the recording, using the Kaplan-Meier estimator which takes account of censored observations, i.e. when the event of interest (wakefulness) was not captured.

Results 14/54 (26%) infants cycled through wakefulness during the recording: durations ranged from 2 to 29 minutes and did not deviate significantly from an exponential distribution (D(14)=0.733, p=0.657) (figure 2 left).

There is a sharp increase in the likelihood of transitioning from sleep to wakefulness when recordings increase from 93 to 96 minutes, when the estimated percentage of subjects maintaining sleep falls from 75 to 50%. Nevertheless, 25% of subjects will still be asleep at 121 minutes (figure 2 right).

Discussion In pre-term infants, the durations of awakenings are exponentially distributed, as in neonatal animals.¹ The likelihood of awakening does not increase linearly with recording duration, but is gated after approximately 100 minutes, demonstrating cyclicity. Future work will build on these preliminary data to model how demographic and environmental variables (e.g. necessary painful procedures) influence the neonatal sleep-wake cycle.


  1. Blumberg, et al. PNAS ( 2005)

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.