Publications Haider S, Tsanas A, Batty GD, Reynolds RM, Whalley HC, Cox SR, Marioni RE, Battersby C, Boardman JP. Statistical learning to identify and characterise neurodevelopmental outcomes at 2 years in babies born preterm: model development and validation using population-level data from England and Wales. EBioMedicine. 2025 Jul;117:105811. doi: 10.1016/j.ebiom.2025.105811. Epub 2025 Jun 17. PMID: 40532624; PMCID: PMC12212173. Lay Summary Background informationBabies born preterm have a higher chance of developmental difficulties in early childhood. These difficulties can affect movement, communication, hearing, and vision, and may influence health, learning, and well-being as children grow. However, most research looks at these developmental areas separately, even though many children experience challenges in more than one area at the same time. With growing interest in using large datasets and modern statistical methods, researchers can now look for hidden patterns of development in a more data-driven way. This study uses routinely collected national data from England and Wales to examine how different developmental difficulties cluster together at two years of age in children born very preterm. It also explores which early-life factors during pregnancy, birth, and the neonatal period are associated with these clusters.Research questionsThis study asked whether there are distinct groups of developmental outcomes at age two among children born very preterm, based on parent-reported difficulties in movement, communication, and sensory function. We also examined whether these groups could be replicated in an independent population and what early-life factors best predict them. This matters because clearer groupings could help identify children who may benefit from targeted follow-up and early interventions.FindingsWe identified four distinct groups of developmental outcomes: children developing typically (85%), those with mainly communication difficulties (8%), those with motor difficulties often linked to cerebral palsy (4%), and a small group with multiple co-occurring impairments (3%). These groups were replicated with high accuracy in the Wales cohort and were strongly aligned with clinical diagnoses and standardised assessments. Brain injuries, such as intraventricular haemorrhage and cystic periventricular leukomalacia, were key predictors for motor and multi-impairment groups. In contrast, communication difficulties were more strongly linked to factors such as gestational age, birthweight, sex, and socioeconomic deprivation than to neonatal medical complications.ConclusionsParents, clinicians, early-years teams, teachers, and policymakers could use these findings to identify children who may benefit from earlier, more tailored support, particularly before starting school. The study also shows that routine parent-reported data can help highlight children at higher risk. Further work is needed to track how these early developmental patterns relate to later outcomes at school age, and to understand how early clinical, social, and educational interventions might reduce later disadvantage. This article was published on 2025-08-21