Background: Many people living with coeliac disease remain undiagnosed for years because symptoms differ widely between individuals and may be mistaken for other conditions. Earlier identification of those at increased risk could reduce diagnostic delay and improve long-term health outcomes. This abstract describes a novel artificial intelligence (AI) framework designed to support population-based risk assessment before confirmatory testing.
Methods: A conceptual AI model was developed by mapping recognised risk factors for coeliac disease into a digital decision-support framework. The proposed model combines demographic characteristics, family history, gastrointestinal symptoms, autoimmune conditions, nutritional indicators, and selected laboratory findings to estimate an individual's probability of having coeliac disease. The framework is intended to support, rather than replace, clinical judgement by helping healthcare providers identify people who may require further investigation.
Results: The proposed framework demonstrates how AI could strengthen public health screening by prioritising individuals with the highest predicted risk for referral and diagnostic assessment. Such an approach may improve the efficiency of screening programmes, encourage earlier diagnosis, and reduce unnecessary investigations among low-risk populations.
Conclusion: AI-supported risk prediction offers a practical opportunity to improve the early detection of coeliac disease within routine healthcare. Future studies should evaluate the framework using real-world clinical data to determine its accuracy, usability, and impact on patient care before wider implementation.