Abstract Background While current IBD research focuses heavily on therapeutic remission, the interception of disease prior to symptom onset represents the next frontier. However, due to the low population prevalence of IBD, precise stratification is fundamentally required to identify at-risk individuals – first to run successful prediction trials and, next, to deploy these preventative measures at scale. Despite advances mapping molecular and environmental factors, the field struggles to combine these into clinical predictive models. We argue that the solution is not a better single test, but a hierarchical stratification strategy. Methods We propose a conceptual hierarchical screening framework (Figure 1), contrasting it against traditional single-timepoint testing. Building on shared immunological pathways across immune-mediated inflammatory diseases (IMIDs), we outline a strategy leveraging ubiquitous data to cast a wide net for broad IMID susceptibility. This initial filter serves as a gateway to increasingly specialised molecular profiling to detect broader signals relevant to IMIDs (e.g., rheumatoid arthritis or IBD). While screening layers are not completely independent, we propose combining static background risk with dynamic molecular signals for complementarity. This multi-disease utility justifies the use of higher-cost profiling before applying high-specificity, disease-exclusive biomarkers. Results Given the ∼1% lifetime risk of IBD, capturing a clinically meaningful proportion (∼30%) of future cases with high precision (80%) would require a virtually unattainable single-timepoint predictor accuracy (AUC 0.98) 1. We hypothesise that the solution lies in a hierarchical funnel. Our analytic modelling suggests that each screening layer effectively reduces the candidate pool by ∼80–90%, exponentially enriching the cohort without requiring perfect accuracy at any single step. While individual layers may possess only modest discriminatory power (AUC ∼ 0.7-0.9), their sequential application isolates a high-risk tail. In this hyper-enriched subset, the model captures a sizeable minority of cases with high positive predictive value, justifying high-cost follow-up and interventions. While current prevention trials rely on first-degree relatives, this framework provides a roadmap for scaling prevention to the sporadic IBD population. Conclusion Future screening strategies must move from a ‘needle in a haystack’ search to a hierarchical filter. By grounding prediction in routine multi-disease data before deploying specialised testing, we can efficiently stratify the high-risk tail. Developing these stratification pipelines in parallel with current prevention trials is critical to ensuring we can effectively scale interventions once validated. References: 1.Jostins L, Barrett JC. Genetic risk prediction in complex disease. Hum Mol Genet. 2011;20(R2):R182-R188. Conflict of interest: Sazonovs, Aleksejs: No conflict of interest
A Sazonovs (Thu,) studied this question.