Development and External Validation of a Machine Learning Model for Pre-Endoscopic Prediction of Montreal Disease Extent (E1/E2/E3) in Ulcerative Colitis Using Symptoms, Signs, and Laboratory Tests: A Multicenter Retrospective Observational Study

Development and External Validation of a Machine Learning Model for Pre-Endoscopic Prediction of Montreal Disease Extent (E1/E2/E3) in Ulcerative Colitis

This multicenter retrospective observational study aims to develop and externally validate a machine learning model that predicts Montreal ulcerative colitis (UC) disease extent (E1: limited/proctitis; E2: left-sided; E3: extensive) using pre-endoscopic clinical information, including symptoms, signs, and laboratory tests. The model is intended to assist clinical assessment before endoscopic confirmation and is not designed to replace colonoscopy or histopathology. Data from development centers (Centers A and B) will be used for model development with nested cross-validation; data from independent external centers (Centers C and D) will be used for external validation only.

Study Overview

• Status: Active, not recruiting
• Conditions: Ulcerative Colitis (Disorder)

• Study Type: Observational
• Enrollment (Estimated): 1500

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

Adults with UC from four hospitals in China (2020-2025). Development cohort: Centers A (Dongfang Hospital, BUCM) and B (Dongzhimen Hospital, BUCM). External validation cohort: Centers C (BUCM Third Affiliated Hospital) and D (Yantai Hospital of Traditional Chinese Medicine).

Description

Inclusion Criteria:

1. Confirmed diagnosis of ulcerative colitis by endoscopy ± histopathology.
2. Montreal disease extent classifiable as E1 (limited), E2 (intermediate/left-sided), or E3 (extensive) and mapped to study labels 1/2/3.
3. Pre-endoscopic baseline data available: demographics, symptoms, signs, and laboratory tests used as model predictors.
4. Predictors collected before or independent of endoscopic findings used for the outcome label (endoscopic extent not used as input).
5. One index visit per patient (duplicate/non-index visits excluded).

Exclusion Criteria:

1. Non-UC diagnosis or Montreal extent not assignable.
2. Missing patient identifier/linkage or unlabelable outcome.
3. Incomplete endoscopic gold standard for Montreal extent classification.
4. Duplicate or non-index visits.
5. Critical predictor data unavailable and not handled by the prespecified modeling pipeline

Study Plan

How is the study designed?

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Time Frame
Macro one-vs-rest area under the receiver operating characteristic curve (macro AUC-OVR) for three-class Montreal extent prediction (E1 vs E2 vs E3) in the external validation cohort (n=247).	At index visit / endoscopic assessment

Collaborators and Investigators

• Sponsor: Beijing University of Chinese Medicine
• Collaborators: Dongzhimen Hospital, Beijing; Dongfang Hospital Beijing University of Chinese Medicine; Beijing University of Chinese Medicine Third Affiliated Hospital; Yantai Penglai Traditional Chinese Medicine Hospital

Study record dates

Study Major Dates

• Study Start (Actual): November 20, 2015
• Primary Completion (Estimated): August 1, 2026
• Study Completion (Estimated): September 1, 2026

Study Registration Dates

• First Submitted: July 19, 2026
• First Submitted That Met QC Criteria: July 19, 2026
• First Posted (Actual): July 23, 2026

Study Record Updates

• Last Update Posted (Actual): July 23, 2026
• Last Update Submitted That Met QC Criteria: July 19, 2026
• Last Verified: July 1, 2026

More Information

Terms related to this study

• Additional Relevant MeSH Terms: Intestinal Diseases; Digestive System Diseases; Gastrointestinal Diseases; Colonic Diseases; Gastroenteritis; Inflammatory Bowel Diseases; Colitis; Colitis, Ulcerative
• Other Study ID Numbers: 2026BZYLL0603

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO
• IPD Plan Description: Due to the restrictions of the informed consent form approved by the Institutional Review Board (IRB), patients were only consented for the use of their data in this specific clinical prediction model study. Broad data sharing was not covered in the original ethical approval, and obtaining re-consent from all 247 participants is not feasible.

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No