Delphi2M AI Forecasts Risk for 1,000+ Diseases Years Ahead

Scientists in Europe have unveiled an artificial intelligence model that can estimate an individual’s risk of developing more than 1,000 diseases years in advance, researchers said in a study published in Nature on Wednesday. The generative AI, code named Delphi2M, was built by teams at the European Molecular Biology Laboratory, the German Cancer Research Centre and the University of Copenhagen and was trained and validated on anonymized health records from the United Kingdom and Denmark.

Delphi2M uses a transformer-based architecture similar to large language models but adapted to medical histories. The model treats diagnoses, demographic details and lifestyle factors such as body mass index, smoking and alcohol use as tokens and learns patterns in the order and timing of medical events to forecast what might occur next. According to the authors, it was trained on about 400,000 participants from the UK Biobank and tested against roughly 1.9 to 2 million records from Denmark, amounting to nearly 2.3 million patient records across the two systems. The system produces risk estimates as rates over time, which researchers liken to a weather forecast.