Multi-center Prospective Validation of Real-Time AI-EGGIM System

Condition(s) studied

Investigational drug(s) / intervention(s)

AI-EGGIM Assessment: During image-enhanced endoscopy, the AI-EGGIM system will assign grades of 0, 1, or 2 to each of five standard gastric regions and calculate a total score from 0 to 10.

Study summary

Gastric intestinal metaplasia (GIM) is a precancerous condition associated with an increased risk of gastric cancer. The endoscopic grading of gastric intestinal metaplasia (EGGIM) score evaluates the extent of GIM across five standard gastric regions. The investigators developed a real-time artificial intelligence-assisted EGGIM (AI-EGGIM) system that automatically assigns an intestinal metaplasia grade to each region and calculates a total EGGIM score.

This prospective multi-center validation study will consecutively enroll patients undergoing upper gastrointestinal endoscopy. The AI-EGGIM system will generate real-time grades for the five standard gastric regions. Regardless of endoscopic appearance or AI output, every patient will undergo five mapping biopsies, with one biopsy obtained from each standard region. Histopathological operative link on gastric intestinal metaplasia (OLGIM) staging will serve as the reference standard. The primary objective is to estimate the performance of an AI-EGGIM score ≥ 5 for identifying patients at high risk of gastric cancer.

Eligibility

Primary outcome measure(s)

• Sensitivity of AI-EGGIM ≥ 5 for Identifying Patients at High Risk of Gastric Cancer — From the endoscopy examination to completion of histopathological assessment, up to 14 days
  Sensitivity will be calculated as the proportion of participants with histopathologically confirmed OLGIM stages III/IV who have an AI-EGGIM score ≥ 5: true positives divided by the sum of true positives and false negatives.

Trial sites (3)

More Qilu Hospital of Shandong University trials in China

Other trials for the same condition