Impact of AI-Assisted Endoscopy on Secondary Review and Referral Decisions in Gastric Cancer Screening: A Multicenter Real-World Study - Scorecard - MDSpire
Clinical Scorecard: Effects of AI-Assisted Endoscopy on Secondary Image Review and Referral Decisions in Gastric Cancer Screening: A Multicenter Real-World Study
At a Glance
Category
Detail
Condition
Population-based mobile gastric cancer screening using esophagogastroduodenoscopy (EGD)
Key Mechanisms
During the AI period, CAD EYE provided real-time visual alerts to support primary interpretation. The primary endoscopist assigned the interpretation and referral category, followed by independent review of recorded images and reports before referral was finalized.
Target Population
Consecutive examinations in a mobile population-based screening EGD program in Gifu Prefecture, Japan; 5,013 examinations were included.
Care Setting
Multicenter mobile endoscopy screening program in Gifu Prefecture, Japan, with structured secondary review.
Key Highlights
Retrospective before–after study comparing 2,021 examinations in 2021–2022 with 2,992 examinations in 2023–2025.
During the AI period, primary interpretation incorporated CAD EYE, which provided real-time visual alerts for regions suspected to represent gastric neoplastic lesions or esophageal squamous cell carcinoma.
CAD EYE was not intended to identify every upper gastrointestinal lesion and did not automatically produce diagnostic classifications or referral determinations.
The primary endoscopist assigned the initial interpretation and referral category before independent secondary review.
Guideline-Based Recommendations
Diagnosis
The source describes a screening workflow in which the primary endoscopist interprets the examination and an independent endoscopist reviews recorded images and reports before referral is finalized.
Management
During the AI period, CAD EYE provided real-time visual alerts as support for primary interpretation; the primary endoscopist retained responsibility for interpretation and referral categorization.
Monitoring & Follow-up
The program's structured secondary review assessed newly detected findings, diagnostic changes, atrophy reassessments, cancer-related changes, and referral recategorizations.
Risks
The study had no external reference standard to determine whether secondary-review modifications or referral decisions were diagnostically correct or clinically appropriate.
The AI platform was not intended to identify every upper gastrointestinal lesion.
Patient & Prescribing Data
Consecutive mobile screening EGD examinations in Gifu Prefecture, Japan; municipalities set target screening ages according to their policies. Examinations without complete image documentation or clinical data were excluded.
This was an evaluation of screening workflow, not a treatment study. Screening used transnasal endoscopy; routine sedation and biopsy sampling were not part of the protocol.
