AI-assisted rapid review of determinants of childhood obesity: a novel approach to evidence synthesis

Childhood and adolescent obesity remains a critical global health issue, with complex, multifaceted causes spanning individual, family, community, and societal levels. This study aimed to identify behavioural and broader determinants of childhood obesity through a review using a novel AI-assisted methodology. A literature search was conducted in Medline, PsycINFO, and PubMed to identify reviews published between 1 January 2007 and 24 April 2025 that (1) involved children or adolescents aged 0–18 years, and (2) investigated determinants of obesity (e.g. screen time, physical activity). A total of 362 systematic review articles were included. Traditional systematic review methods were combined with large language models to expedite eligibility screening, full-text data extraction, and thematic synthesis. The review identified a wide range of determinants. Behavioural determinants included unhealthy dietary patterns, low levels of physical activity, excessive screen time, and insufficient sleep. These behaviours were often shaped by social and environmental determinants including family routines, parental influence, peer norms, school policies, and broader structural factors such as socioeconomic status, food marketing, and neighbourhood environments. Environmental constraints, such as limited access to green spaces and affordable healthy food, were especially influential in low-income contexts. AI-assisted methods enabled rapid and structured extraction of complex data across a large, heterogeneous body of literature, improving both the speed and consistency of synthesis while maintaining rigour through human oversight. AI tools supported the identification of recurring themes, classification of determinants, and cross-context comparisons. This project demonstrates the potential of AI to enhance the efficiency of evidence synthesis in addressing complex public health challenges. The findings underscore the complex, multi-layered nature of obesity determinants and support the need for multi-factorial interventions across behavioural, social and structural levels.