Around 8,000 babies are born very early – before 32 weeks of pregnancy – each year in the UK.[1-3] Sadly, more than one in three of these tiny babies will develop a lung condition called bronchopulmonary dysplasia (BPD) or lose their lives within the first few weeks after birth. Dr T’ng Chang Kwok of the University of Nottingham is using artificial intelligence (AI) to analyse routine data in electronic medical records to identify which very preterm babies are most at risk of dying or developing severe BPD. He hopes to develop a new tool that can guide personalised treatment decisions, improving outcomes for these vulnerable babies and their families.
How are children’s lives affected now?
Babies born very prematurely can develop BPD because their lungs have not fully developed in the womb. They will usually need oxygen therapy via a ventilator to support their breathing and, while this treatment is life-saving, it can sometimes cause damage to their tiny lungs.
“While most affected babies get better as they grow older, some will develop breathing difficulties – such as coughing, wheezing or breathlessness – later in childhood,” says Dr Kwok, “Those who are most severely affected with BPD may need long-term oxygen therapy at home – and they may also experience learning difficulties.”
Doctors can prescribe different medicines, such as steroids, to help prevent BPD in preterm babies. But these treatments can have serious side effects, such as cerebral palsy – and doctors must balance the potential risks and benefits when making decisions.
“Delaying treatment can make it less effective – but the early treatment of all babies would mean some are exposed to unnecessary risks,” says Dr Kwok. “Predicting who is most likely to benefit from treatment and when would give each child the best chance of a successful outcome.”
How could this research help?
“We aim to develop a new AI-based tool that can support clinical decision-making for premature babies who are at high risk of dying or developing BPD,” says Dr Kwok.
The researchers will analyse data within electronic medical records collected from many thousands of very preterm babies – which are routinely recorded on an NHS system called BadgerNet.
“By accurately recognising complex patterns within these data, AI could identify which very preterm babies are likely to benefit from treatment – and pinpoint the best time to start,” says Dr Kwok. “
This new tool – called PRIOR-AI (PRemature Infant Outcome Risk – Artificial Intelligence) – would be made freely available and integrated into BadgerNet so that doctors across the UK can use it immediately to tailor treatment plans for very preterm babies.
- Office for National Statistics (England and Wales). Birth Characteristics, 2020 edition, Table 8: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/livebirths/datasets/birthcharacteristicsinenglandandwales [website accessed 25 May 2022]
- Public Health Scotland, Births in Scottish hospitals: https://publichealthscotland.scot/publications/births-in-scottish-hospitals/births-in-scottish-hospitals-year-ending-31-march-2021/ [website accessed 28 June 2022]
- Northern Ireland Statistics and Research Agency, Registrar General Annual Report 2020. https://www.nisra.gov.uk/system/files/statistics/RG%20Annual%20Report%202020%20Accessible.pdf [website accessed 25 May 2022]
- Royal College of Paediatrics and Child Health, National Neonatal Audit Programme Annual Report on 2020 data: https://www.rcpch.ac.uk/resources/national-neonatal-audit-programme-annual-report-2020 [website accessed 25 May 2022]
Research Training Fellowships:
Each year, Action Medical Research awards these prestigious grants to help the brightest and best doctors and scientists develop their career in medical research.
|Project Leader||Dr T’ng C Kwok, MRCPCH BMBS BMedSci|
|Location||Academic Child Health, School of Medicine, University of Nottingham|
Professor Don Sharkey, BMedSci BMBS PhD FRCPCH
Professor Carol Coupland, PhD
Professor Jon Garibaldi, PhD
|Other Locations||School of Computer Science, University of Nottingham|
|Grant Code (GN number)||GN2941|