AI to personalise treatment for leading cause of gastric cancer

Researchers have developed an artificial intelligence system that personalises treatment for Helicobacter pylori.

"This study shows how AI can harness real-world clinical data to retrospectively optimise Helicobacter pylori eradication strategies - paving the way toward a new era of precision, data-driven medicine." Dr Kirill Veselkov Corresponding author

Helicobacter pylori (H. pylori) is the most common carcinogenic pathogen globally and the leading cause of gastric cancer.

The study, published in Nature Communications and led by Dr Kirill Veselkov from Imperial’s Department of Surgery and Cancer, details the development of a reinforcement learning–based AI Clinician system designed to tailor treatment selection for H. pylori.

The AI model analyses a wide range of patient factors—including age, sex, antibiotic allergies, geographic region, symptoms, and concurrent medications—to recommend the most effective first line therapy. In a retrospective study of more than 38,000 patients, with external validation in over 7,000 additional cases, the AI Clinician system improved eradication success by six percent compared with standard clinician selected treatments.

Supported by UKRI and the EU’s AIDA project, this innovation has the potential to significantly advance global cancer prevention by optimising therapies for a pathogen that affects nearly half the world’s population.

Speaking about the research, Dr Kirill Veselkov said: "This study shows how AI can harness real-world clinical data to retrospectively optimise Helicobacter pylori eradication strategies - paving the way toward a new era of precision, data-driven medicine."

Kyle Higgins, Olga P. Nyssen, Joshua Southern, Ivan Laponogov, AIDA CONSORTIUM, Dennis Veselkov, Javier P. Gisbert, Tania Fleitas Kanonnikoff & Kirill Veselkov. The Helicobacter pylori AI-clinician harnesses artificial intelligence to personalise H. pylori treatment recommendations. Nat Commun 16, 6472 (2025). https://doi.org/10.1038/s41467-025-61329-5