AI-based quality auditing of esophagogastroduodenoscopy

Source article: A Video-Based Artificial Intelligence Platform for Measuring Esophagogastroduodenoscopy Duration and Suggesting Real-Time Landmark Documentation

Background and Study Aims The European Society of Gastrointestinal Endoscopy (ESGE) sets key quality indicators for esophagogastroduodenoscopy (EGD), including a minimum procedural time of 7 min and photodocumentation of landmarks. We evaluated AI-derived procedural times, physician variations, and the impact of AI suggestions on landmark documentation rates. Methods This retrospective multicenter study included 78,336 EGDs performed by 213 physicians (01.01.2024-31.12.2025). A video-based AI platform (Theator I…

A Video-Based Artificial Intelligence Platform for Measuring Esophagogastroduodenoscopy Duration and Suggesting Real-Time Landmark Documentation
"Endoscopy" by Yuya Tamai is licensed under CC BY 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/2.0/.
Trace impact readingContested
P 70The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.

Both sides are scored from claims and sources, not community votes.

G 69The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.

In brief

Researchers analyzed 78,336 upper endoscopies by 213 physicians over 2024 to 2025 using a video-based AI platform. The system measured procedure duration, offered real-time suggestions for photographing anatomical landmarks, and tracked whether those photos were documented. Two blinded experts checked 500 procedures to validate the timing.

The findings show that automated, unattended measurement can audit endoscopy quality at scale and reveal wide variation among physicians. Procedure times often fell short of the European benchmark, which raises concerns about thoroughness. The study is retrospective, so the documentation gains are reported as an association rather than proven cause.

Main points

  1. The study covered 78,336 EGDs by 213 physicians from January 2024 to December 2025.
  2. Median procedure time was 3.30 minutes, with an interquartile range of 2.27 to 4.65.
  3. Among 203 physicians with at least 50 procedures, average times ranged from 1.35 to 7.56 minutes, a 5.6-fold difference.
  4. Zero-landmark procedures fell from 27.5% to 15.8% with AI suggestions, and the association persisted after restratifying by biopsy-free time.

The gain

AI timestamps matched blinded expert review in 98.5% of checked cases, and AI landmark suggestions cut procedures with zero documented landmarks by 42.6%.

The problem

Measured procedure times fell far below the ESGE 7-minute benchmark, with a median of 3.30 minutes and only 6.1% of procedures meeting the threshold.

The rundown

ESGE quality indicators call for a minimum procedural time of 7 minutes and photodocumentation of landmarks. The study tested whether an AI platform from Theator could measure these indicators automatically from video.

Validation used two blinded experts reviewing 500 procedures, with agreement defined as within 2 seconds. AI timestamps agreed with them in 98.5% of cases, with a 95% confidence interval of 97.1 to 99.3%.

Only 6.1% of procedures reached the 7-minute threshold, and individual physicians' typical times varied 5.6-fold. The authors conclude that automated measurement offers a scalable route to auditing upper endoscopy quality.

Sources

  1. Peer-reviewedEndoscopy International Open2026-10-13

The debate