Skip to content

Understanding pain and discomfort in preterm babies using artificial intelligence

Preterm Enhanced Automated Capture Of Comfort Knowledge: the Peacock study

Status
Recruiting
Phases
Unknown
Study type
Observational
Source
ISRCTN
Registry ID
ISRCTN11335900
Enrollment
50
Registered
2022-07-19
Start date
2022-08-01
Completion date
Unknown
Last updated
2022-09-05

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Preterm infant discomfort during intensive care Neonatal Diseases

Interventions

The study uses computer learning to evaluate a video of the baby during routine procedures and learns as more cases are observed. No changes to routine care are involved. Participatin

Sponsors

Newcastle upon Tyne Hospitals NHS Foundation Trust
Lead Sponsor

Eligibility

Sex/Gender
All

Inclusion criteria

Inclusion criteria: 1. Born at <36 weeks gestation 2. Medically stable 3. Signed parental consent

Exclusion criteria

Exclusion criteria: 1. Infants with significant brain, spine or facial congenital abnormality 2. Parents unwilling to provide consent 3. Infants with postmenstrual age >36 weeks

Design outcomes

Primary

MeasureTime frame
How well the final learning model performs, assessed using machine learning metrics (confusion matrix, accuracy, precision, recall/sensitivity, F1 score, specificity and area under the curve) at the time of the video recording

Secondary

MeasureTime frame
1. What signal or mixture of signals is most informative of preterm babies’ state (face, body, sound and physiological data) at the time of the video 2. What algorithms are best suited for each data stream and why 3. Which methods are best for combining different data streams to make the most accurate estimation of preterm babies’ comfort levels 4. Are any of these factors different for some babies e.g. the most immature, those with ventilator devices on their faces 5. Can these models detect prolonged pain (e.g., chronic pain post-surgery), can they distinguish this from acute procedural pain 6. What medical and contextual factors affect behavioural responses to pain 7. How many recordings are required to achieve the best model performance All will be measured at the time of the video recording and assessed using performance metrics for machine learning outlined in the primary outcome measure

Countries

England, United Kingdom

Outcome results

None listed

Source: ISRCTN (via WHO ICTRP) · Data processed: Feb 4, 2026