This AI calculates the risk that a patient has of contracting 1000 diseases within ten years.

This article is taken from the monthly magazine Sciences et Avenir n°945, dated November 2025.

Using algorithmic concepts similar to those used in large language models such as ChatGPT, a European team presented in the journal Nature a generative AI capable of calculating the individual risk of contracting more than 1000 diseases in the next decade.

Medical events and lifestyles examined through the lens of AI

Researchers from the European Molecular Biology Laboratory (EMBL), the German Cancer Research Center, and the University of Copenhagen (Denmark) trained their Delphi-2M model using data from 400,000 patients from the UK Biobank study and 1.9 million individuals from the Danish National Patient Register. The AI searches for medical events in the patient's history (date of diagnosis of specific diseases) and lifestyle factors such as age and sex.

Read alsoDepression, bipolar disorder: could AI-powered facial emotion analysis prevent relapses?

"Medical events often follow predictable patterns."

"Medical events often follow predictable patterns," note Tomas Fitzgerald, co-author of the study. Our model learns these patterns to predict future health outcomes. “