Artificial Intelligence Model for Diagnosing Helicobacter Pylori Infection and Premalignant Gastric Conditions

Strategic Insights -

A team of researchers from National Taiwan University Hospital has developed a rural-to-center artificial intelligence (AI) model to automatically evaluate upper endoscopic images captured during routine clinical practice. The model was trained on endoscopic images from a rural hospital on Matsu Islands and a tertiary center across Taiwan Strait. The AI model was able to detect Helicobacter pylori infection, atrophic gastritis, and intestinal metaplasia with high accuracy.

Key Takeaways:

  • The AI model was trained on 13,118 endoscopic images from 2012 to 2022, with a diagnostic accuracy of 92.8% for H. pylori, 88.6% for atrophic gastritis, and 88.0% for intestinal metaplasia.
  • The model was implemented in rural communities from 2023 to 2024, with 3518 residents undergoing 13C urea breath testing or histological assessment.
  • The implementation-phase diagnostic accuracies were 91.3% for H. pylori, 79.9% for atrophic gastritis, and 63.4% for intestinal metaplasia.
  • The study concluded that AI enables frontline physicians in resource-limited settings to rapidly assess gastric health using routinely captured endoscopic images.
  • The research was funded by the National Taiwan University Hospital and the National Science and Technology Council.

Statistics:

  • 13,118 endoscopic images were used to train the AI model from 2012 to 2022.
  • 3518 residents were invited to undergo 13C urea breath testing or histological assessment in rural communities.
  • The implementation-phase diagnostic accuracies were 91.3% for H. pylori, 79.9% for atrophic gastritis, and 63.4% for intestinal metaplasia.

Sources:

  • A Rural-to-Center Artificial Intelligence Model for Diagnosing Helicobacter pylori Infection and Premalignant Gastric Conditions Using Endoscopy Images Captured in Routine Practice. Endoscopy, 2025.
  • Medical Devices & Surgical Technology Week. November 2, 2025; p 1382.
  • Georg Thieme Verlag Kg, Rudigerstr 14, D-70469 Stuttgart, Germany (publisher of Endoscopy)
  • Yen-Ning Hsu, Center of Intelligent Healthcare, National Taiwan University Hospital, Taipei City, Taiwan (corresponding author)