A practical AI model for endoscopic diagnosis of gastric premalignant lesions and its application to predicting metachronous recurrence.

Perform individual gastric risk assessment and staging of precancerous conditions at first-time gastroscopy, irrespective of the patient's country of origin. Use high quality endoscopy with virtual chromoendoscopy (VCE), after proper training, for screening, diagnosis and staging of atrophy and intestinal metaplasia and after endoscopic therapy; VCE guides the biopsy site, with random biopsies taken only when there are no endoscopically suspected changes. Take biopsies from at least two topographic sites, 2 from the antrum/incisura and 2 from the corpus, VCE-guided, in two separate, clearly labeled vials; an additional incisura biopsy is optional, not required. Stage with validated endoscopic classifications of atrophy (e.g. Kimura-Takemoto) or intestinal metaplasia (e.g. EGGIM) to stratify gastric cancer risk (suggestion, weaker than the biopsy recommendation). Follow up patients with extensive endoscopic changes (Kimura C3+ or EGGIM 5+) OR an advanced histological stage that reaches OLGA/OLGIM III/IV, severe atrophy or intestinal metaplasia and/or significant changes in both antrum and corpus rising to stage III/IV, not any intestinal metaplasia or any antrum-plus-corpus involvement on its own, with high quality endoscopy every 3 years, irrespective of country of origin (strong recommendation, moderate quality). Propose NO surveillance for patients with mild to moderate atrophy or intestinal metaplasia restricted to the antrum, in the absence of endoscopic signs of extensive lesions or other risk factors (family history of gastric cancer, incomplete intestinal metaplasia, persistent H. pylori infection), the guideline states this group constitutes most individuals found in clinical practice. Eradicate H. pylori in all patients with precancerous conditions and after endoscopic or surgical therapy. Discontinue or do not start gastric cancer screening or surveillance in asymptomatic individuals over 80, and weigh comorbidities when planning treatment of superficial lesions. Advise smoking cessation; low-dose daily aspirin may be considered for gastric cancer prevention only in selected individuals already at high cardiovascular risk. For visible neoplasia, use ESD for differentiated lesions clinically staged as dysplastic (low or high grade) or intramucosal carcinoma (any size if not ulcerated, or ≤30 mm if ulcerated), with EMR an alternative for Paris 0-IIa lesions ≤10 mm with low likelihood of malignancy. At the population level, endoscopic screening is suggested every 2-3 years in high-risk regions (age-standardized rate >20 per 100,000 person-years), every 5 years in intermediate-risk regions (ASR 10-20) only if cost-effectiveness has been proven, and not at all in low-risk regions (ASR <10).