Cure8 research brief
Earlier identification of people likely to develop IBD could shorten diagnostic delays that are linked to worse outcomes, and an automated EHR model might help target who should get earlier evaluation.
Clinicians who evaluate patients with gastrointestinal symptoms, researchers in AI/healthcare and IBD biology, health system leaders planning EHR-based case-finding, and patients and advocates interested in reducing diagnostic delay.
The study developed a state-space model (Mamba) using pre-diagnostic EHR features while excluding elements of the diagnostic workup. On held-out UCSF data and an external safety-net hospital (ZSFG), the model showed high AUROC at 6 months and estimated median lead times measured in many months.
The authors also simulated a lower-burden deployment using diagnosis-code–only inputs and estimated how many alerts per month would need review to find additional cases.
The report is a preprint (not yet peer reviewed) and presents proof-of-concept results suggesting EHR-based screening could identify a population for prospective studies of earlier IBD biology and potentially reduce diagnostic delay. It does not test whether earlier identification improves clinical outcomes.
Practical next steps noted by the authors include prospective evaluation and assessment of real-world deployment challenges (e.g., alert burden, generalizability to diverse health systems).
This is a medRxiv preprint (not peer reviewed) reporting retrospective development and external validation using two US health systems. Model performance metrics are promising but require prospective testing to show clinical benefit and to evaluate implementation trade-offs such as false positives and review workload.
Review the original publication for the complete reporting, methods, and context.
This Cure8 brief is based on source text from the linked article. Cure8 is informational only and is not a substitute for professional medical advice, diagnosis, or treatment.
Automated analysis of full-length endoscopy videos could standardise assessment of UC activity, identify spatial patterns of inflammation, and better estimate histological healing—information that may influence treatment...
A validated pre-endoscopic model could help estimate UC disease extent before colonoscopy, aiding clinical triage and planning without replacing endoscopy.
Researchers are exploring ways to prioritize who needs colonoscopy sooner; adding TCM features to prediction models might change triage performance. Patients and clinicians interested in pre-endoscopic risk tools or...
