A groundbreaking artificial intelligence model, Delphi-2M, can forecast a person’s risk of developing more than 1,000 diseases — in some cases providing predictions up to 20 years ahead. The study, published today in Nature, highlights the potential of AI to revolutionize preventive medicine.
Developed by researchers including Moritz Gerstung at the German Cancer Research Center, Delphi-2M uses a modified large language model (LLM) to analyze health records, lifestyle factors, and demographic information such as age, sex, and body mass index. The AI estimates the likelihood of conditions ranging from cancers and skin diseases to immune disorders.
“Most existing AI tools can only predict a single disease,” said Gerstung. “A healthcare professional would have to run dozens of them to deliver a comprehensive answer. Delphi-2M can do all of this at once.”
The model was trained on data from 400,000 participants in the UK Biobank, a long-term biomedical study, and tested on 1.9 million people from Denmark’s National Patient Registry. For most diseases, Delphi-2M’s predictions matched or exceeded the accuracy of traditional single-disease models, and outperformed biomarker-based algorithms that rely on specific molecular measurements.
Stefan Feuerriegel, a computer scientist at Ludwig Maximilian University of Munich, called the tool’s capabilities “astonishing,” noting that it can generate full future health trajectories for individuals. Delphi-2M was particularly accurate for conditions with predictable progressions, such as certain types of cancer.
While promising, the model has limitations. It currently uses only first-occurrence disease data from the UK Biobank, which may not capture the full trajectory of a person’s medical history. Researchers plan to expand testing to datasets from other countries to improve accuracy and generalizability.
Delphi-2M represents a significant advance in multi-disease prediction and early-warning systems, offering clinicians the potential to identify high-risk individuals and deploy preventive interventions long before symptoms appear.
