Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1937
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-19T22:02:17.694195Z
status: published
lens: g_space
sector: health
headline: A practical AI model for endoscopic diagnosis of gastric premalignant lesions and its application to predicting metachronous recurrence
dek: Background/aims To evaluate the diagnostic performance of an artificial intelligence (AI)-based optical system for chronic atrophic gastritis (CAG) and intestinal metaplasia (IM), and its ability to predict metachronous recurrence. Methods We analyzed 72,794 endoscopic images from 5,000 patients undergoing esophagogastroduodenoscopy with biopsy-confirmed CAG/IM. A convolutional neural network model was developed and compared with experts (n=16) and novices (n=15). The association between AI-predicted IM extent a…
gain_title: A neural network classified chronic atrophic gastritis more accurately than both novice and expert endoscopists in a single-center study, and its predicted intestinal metaplasia extent showed a dose-response trend with recurrence.
problem_title: (none)
trace_subject: (none)
gain_reading: A neural network classified chronic atrophic gastritis more accurately than both novice and expert endoscopists in a single-center study, and its predicted intestinal metaplasia extent showed a dose-response trend with recurrence.
gain_evidence: a significant dose-response relationship with recurrence risk was observed (P trend =0.043)
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers built a convolutional neural network trained on 72,794 endoscopic images from 5,000 patients with biopsy-confirmed chronic atrophic gastritis or intestinal metaplasia. They compared its performance with 16 experts and 15 novices, then tested whether the AI-estimated extent of intestinal metaplasia related to metachronous recurrence in 458 patients after endoscopic resection of early gastric cancer.

The work suggests AI could help endoscopists recognize stomach changes that raise cancer risk and could aid in judging who is more likely to have recurrence. The authors describe the evidence as coming from a single center, so it is not yet known whether the model performs as well in other hospitals or populations before clinical use.
limitation: The model was tested in a single-center study and the authors state it needs multicenter external validation before clinical adoption; the individual hazard ratios for recurrence were not statistically significant.
tag: Evidence-backed gain
key_points: For chronic atrophic gastritis, AUCs were 0.885 in the antrum and 0.901 in the body. | For intestinal metaplasia, accuracies were 0.759 and 0.723, with AUCs of 0.835 and 0.744. | The model beat novices for intestinal metaplasia diagnosis, 0.875 versus 0.781 accuracy. | Hazard ratios for AI-predicted intestinal metaplasia extent were not individually significant, but the trend test was.
rundown: The dataset covered patients undergoing esophagogastroduodenoscopy with biopsy-confirmed disease. The model was evaluated at two sites in the stomach, the antrum and the body, with differing accuracies for each lesion type.

In the clinician comparison, the AI outperformed both novices and experts for atrophic gastritis. For intestinal metaplasia, the reported advantage was over novices only.

For recurrence, Cox regression was applied to 458 patients after endoscopic resection of early gastric cancer. Hazard ratios were 2.11 for antral metaplasia only and 4.27 for metaplasia extending to the body, with wide confidence intervals that included 1, though the trend across extent categories was significant.
sources:
- peer_reviewed | Digestive and Liver Disease | https://doi.org/10.1016/j.dld.2026.10.005 | 2026-10-16
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

231d368ec51d1a4c4b8fd1c610c8df99d42c93f2d0d797b37a45f90d04923038

previous

0000000000000000000000000000000000000000000000000000000000000000