Introduction

In ulcerative colitis (UC) clinical trials, endoscopic disease activity is commonly assessed via central reading of the overall Mayo Endoscopic Score (MES). Segmental MES can provide a more detailed evaluation of mucosal response, but it is limited by high inter-rater variability in both anatomical landmark localization and disease activity grading. Automatic MES grading can assist reducing reader discrepancies and detect local mucosal changes that may not be estimated by global scoring. In this study, we developed and evaluated an artificial intelligence (AI)-based system for automated segmental MES grading enabling disease extent quantification and granular endoscopic remission estimation.

Aims & Methods

In contrast to previous approaches, the proposed AI system introduces a novel hierarchical framework for automated MES grading at the frame-, segment-, and video-level. The system comprises three modules, trained on 5,897 retrospectively acquired, aggregated, and anonymized endoscopic videos. A foundation model (FM), trained on 22 million frames, was developed to extract clinically relevant frames from full-length colonoscopies. A colon segmentation module was built upon the FM to localize three anatomical segments: descending colon (DC), sigmoid colon (SI), and rectum (RM). An automated MES classification module was then trained to perform frame-level MES grading. Hierarchical aggregation of local scores, combined with temporal smoothing and non-informative filtering, enabled both segmental and global MES estimation. For independent testing, 915 colonoscopy videos were annotated for each segment for MES scoring by expert gastroenterologists using a 2+1 reading paradigm. Colon segment reference annotations were automatically extracted from endoscopic videos based on recording software annotations. System performance was evaluated using the Quadratic Weighted Kappa (QWK) for both segmental and overall MES agreement. Classification accuracy in detecting endoscopic granular remission (MES 0, 1 vs 2, 3) was reported for all colon segments.

Results

The AI system demonstrated strong agreement (QWK=0.752 [95% CI: 0.719, 0.782]) with gastroenterologist assessments for overall MES on the test dataset. Good agreement was confirmed for SI (QWK=0.647 [95% CI: 0.602, 0.687]) and RM (QWK=0.687 [95% CI: 0.646, 0.724]), moderate agreement was found for DC (QWK=0.592 [95% CI: 0.541, 0.638]). Frame-level colon segment classification accuracy was 41% for DC, 69% for SI, and 78% for RM; with lower performance in DC linked to high variability in reference annotations. Colon segmentation enabled a 37.72% reduction in mean colonoscopy video time by filtering out-of-body and insertion-phase sections. The system showed very high accuracy in detecting both overall (ACC=90.38%) and segmental (DC: 78.80%; SI: 82.84%; RM: 85.03%) endoscopic remission. Mean accuracy in detecting the number of colon segments per video with active disease was 59.13%.

Conclusion

We introduced a novel AI system for automated, granular MES scoring in ulcerative colitis patients, hierarchically grading colonoscopy videos. Our framework demonstrated strong agreement with gastroenterologists in both overall and segmental endoscopic activity assessments. By integrating continuous scoring with anatomical landmarking, the AI system can offer a more comprehensive view of total disease burden in UC patients which can lead to improved precision of disease estimation and treatment response evaluation in clinical settings.

Disclosure

All authors are employees of Clario Inc., Philadelphia, USA.

Introduction

Placebo and treatment response rates in randomized ulcerative colitis (UC) trials impact clinical trial design and patients required to demonstrate efficacy. Prior efforts to predict endoscopic healing (EH) in UC trial patients using clinical measures have failed to produce meaningful predictive power. Using AI analysis of baseline endoscopy videos from the ustekinumab UNIFI registration Phase 3 trial in moderate to severe UC, we demonstrate robust accuracy to predict 8 week endoscopic healing in ulcerative colitis patients treated with placebo or drug.

Aims & Methods

300 screening endoscopy videos were obtained from UC patients receiving placebo, and 606 screening endoscopy videos were obtained from UC patients receiving ustekinumab (130mg or 6mg/kg) in the UNIFI induction trial. For all videos, visual features were extracted using EndoDINO, an AI foundation model pre-trained on the largest reported endoscopy video dataset. For each patient, 21 baseline clinical measures were also obtained as predictors. These included fecal calprotectin, albumin, C-reactive protein, lactoferrin, Geboes grading, prior biologics failure, prior steroids failure, UCEIS scores, and Mayo scores.
For each of the placebo and drug arms, we trained two binary classification models: one using the AI visual features and the other using the baseline clinical measures as predictors. Each model was trained to predict 8 week EH, defined as an endoscopy subscore of the Mayo score equal to 0 or 1. We performed 5-fold cross validation and reported the average Area Under the Receiver Operating Curve (AUROC).
This study, carried out under YODA Project # 2024-0668, used data obtained from the Yale University Open Data Access Project, which has an agreement with JANSSEN RESEARCH & DEVELOPMENT, L.L.C.. The interpretation and reporting of research using this data are solely the responsibility of the authors and does not necessarily represent the official views of the Yale University Open Data Access Project or JANSSEN RESEARCH & DEVELOPMENT, L.L.C.. The original proposal can be found: https://yoda.yale.edu/data-request/2024-0668/.

Results

For the placebo arm, our best performing model using AI visual features predicted 8 week endoscopic healing with an AUROC of 0.78, compared to 0.70 for our best performing model using the combination of 21 baseline clinical measures.
For the drug arm, our best performing model using AI visual features predicted 8 week endoscopic healing with an AUROC of 0.75, compared to 0.72 for our best performing model using the combination of 21 baseline clinical measures.
For models using baseline clinical measures, table 1 shows the importance factor for select measures across arms, calculated using XGBoost built in feature importance.

Conclusion

Our results demonstrate AI analysis of baseline colonoscopy videos with the EndoDINO AI foundation model provides robust accuracy for predicting 8 week EH in ulcerative colitis patients in both placebo and drug arms. AI placebo response prediction could be used in future trial design as covariate analysis to increase statistical power or reduce necessary sample size.
In future work, we aim to improve predictive power by combining our AI visual features with the most important baseline clinical measures through different modeling techniques.

References

Sands BE, Sandborn WJ, Panaccione R, et al. Ustekinumab as induction and maintenance therapy for ulcerative colitis. N Engl J Med. 2019;381(13):1201–1214. doi:10.1056/NEJMoa1900750

Disclosure

AK, PD, MC, and MS are employees of Virgo Surgical Video Solutions, Inc., which developed technology used in this work.

Introduction

UC-SCALE[1] is an AI algorithm designed to assess endoscopic severity in ulcerative colitis (UC). It was trained on a large subset of Phase III Etrolizumab (ETRO) clinical trial data from 471 sites. In our previous work, UC-SCALE previously demonstrated excellent agreement (quadratic-weighted kappa >0.80) with centrally read Mayo Clinic Endoscopic Subscores (MCES) on an ETRO test set spanning 83 sites. While clinical trial data is highly controlled, its generalizability to real-world settings remained untested. Here, we assess UC-SCALE on a retrospective cohort from the Paris IBD Center, which we refer to as the Real World Dataset (RWD).

Aims & Methods

The RWD contained 276 unique UC patients (mean 1.17 visits/patient) from the Paris IBD center. For each colonoscopy video, two or three local gastroenterologists reviewed the recordings and reached a consensus MCES (0–3) per colonic segment (descending, sigmoid, rectum). Using the same, unmodified UC-SCALE model trained on ETRO, we evaluated both the RWD and ETRO test sets with the identical assessment protocol detailed in [1]. Quadratic-Weighted Kappa (QWK) quantified the agreement between ordinal predictions made by UC-SCALE and human scoring. Area Under the Receiver Operating Characteristic Curve (AUC) was used to measure the discriminative ability for each MCES class.

Results

On these real-world data, UC-SCALE’s section-level predictions closely mirrored local expert scoring. QWK reached ~0.804 - virtually identical to ETRO (~0.802). For class-specific discrimination (MCES 0–3), AUC values exceeded 0.85 in all classes and in several cases surpassed ETRO’s results. The average AUC across all classes approached 0.90 in the RWD, aligning well with performance in trial conditions. As in ETRO, performance was strongest for Mayo 0 and 3, but notably, predictions for intermediate scores (Mayo 1 and 2) were even more accurate in RWD. These results highlight UC-SCALE’s robust generalizability to routine clinical practice.

Conclusion

Applying the original ETRO-trained UC-SCALE model to a real-world UC cohort confirmed its high accuracy and reliability for objective MCES scoring outside of clinical trials. To date, our analysis has focused on predictions at the level of anatomical segments. Future work will expand toward full video-level scoring (closer to clinical workflows) and artificial segments, which offer a more granular topological severity map. Additionally, upcoming analyses will explore correlations between UC-SCALE and key IBD biomarkers, as well as direct comparisons with human scoring.

References

[1] Gutierrez-Becker, Benjamin, et al. "Ulcerative Colitis Severity Classification and Localized Extent (UC-SCALE): An Artificial Intelligence Scoring System for a Spatial Assessment of Disease Severity in Ulcerative Colitis." Journal of Crohn's and Colitis 19.1 (2025): jjae187.

Disclosure

Alvaro Gomariz, Bernhard Stimpel, Benjamin Gutierrez-Becker, Stefan Frässle, Emily Fisher, Steven Levitte and Teresa Karrer are employees of F. Hoffmann-La Roche Ltd.

Introduction

The Tokyo Guideline 2018 recommends laparoscopic cholecystectomy as the gold standard therapy for acute cholecystitis regardless of the severity and the onset of symptoms, as long as the patient is fit for surgery. Approximately 20% of the patients undergo acute cholecystectomy in our region, and in the remainder of the patients, we have to take into consideration the likelihood of recurrent inflammation after the regression with conservative therapy. According to the TG 2018, the drainage of the gallbladder is recommended in grade II and grade III cholecystitis. This can be carried out by interventional radiologist as percutaneous transhepatic gallbladder drainage (PTGBD), by endoscopist as transpapillary gallbladder drainage (TPGBD) or endosonography-guided gallbladder drainage (EUS-GBD).

Aims & Methods

The primary endpoint of our prospective, non-randomized, observational study was to determine the technical (successful stent implantation into the gallbladder) and clinical success (repeated need for drainage and absence of recurrent cholecystitis) of TPGBD in those patients that underwent it from 01.01.2023 at the University of Szeged. Our secondary endpoint was the comparison of the time of hospitalization between the patients who underwent clinically successful TPGBD and those who underwent PTGBD for grade II/III cholecystitis.

Results

Altogether 19 patients had undergone TPGBD, 74% of them were female, the average age was 64,53. Previously 21% had cholecystitis, 47% had cholecystolithiasis, while 16% had choledocholithiasis and 11% of the patients gone under PTGBD. TPGBD was carried out because of grade II cholecystitis in 89% of the patients, and grade III in the rest of them. 63% was considered at least a 3 in the ASA physical status classification. Altogether the technical success rate was 58%, a cholangioscop was used in 37% of the cases. The successful canulation rate of the bile duct was 89%, while the canulation rate of the cystic duct was 68%. There were two cases, where after the canulation of the cystic duct the implantation of the plastic stent was not possible. 3 patients were referred for PTGBD after an unsuccessful TPGBD, the remainder patients with an unsuccessful TPGBD received conservative therapy. After a technically successful TPGBD, only one patient went under PTGBD due to his severe comorbidities, rather than a repeated TPGBD. 6 patients underwent cholecystectomy averagely in 8,5 weeks after the TPGBD, the remainder of patients did not have neither the need for repeated drainage nor any symptoms requiring urgent care due to cholelithiasis. Comparing this group with 11, randomly chosen patients who underwent PTGBD for at least grade II cholecystitis during the period of the study, we detected that the time of hospitalization after the drainage was less (8,2 days vs 13,3 days), and less patients required repeated drainage or had recurrent cholecystitis (1 vs 4).

Conclusion

Although TPGBD imposed a great challenge before the use of cholangioscop in everyday practice, nowadays it is becoming an accepted part of the therapeutic algorithm in the treatment of cholecystitis, where an acute cholecystectomy is not performed. Our results suggest that TPGBD can lead to shortened hospitalization and have less complications than TPGBD. Our findings will serve as a basis for a future extension of the study and for the design of a randomised prospective study with a control group.