AI tool predicts 1,000+ diseases that can afflict one up to 20 years in advance

Scientists have developed Delphi-2M, an AI tool that can predict a person’s risk of developing more than 1,000 diseases up to 20 years in advance by analyzing medical records and lifestyle factors. The large language model represents a significant advancement in predictive healthcare, potentially enabling clinicians to identify high-risk patients and implement preventive measures decades before symptoms appear.

What you should know: Delphi-2M uses a modified version of the same technology that powers ChatGPT to forecast disease risk across multiple conditions simultaneously.

• The model analyzes past medical history along with age, sex, body mass index, and health habits like tobacco and alcohol use to generate comprehensive health predictions.
• Unlike existing tools that typically predict risk for only one disease, Delphi-2M can assess the likelihood of developing 1,258 different conditions in a single analysis.
• Researchers trained the model on data from 400,000 participants in the UK Biobank, a long-term biomedical study that tracks health outcomes over time.

Key performance metrics: The AI tool matched or exceeded the accuracy of current single-disease prediction models across most conditions.

• Delphi-2M performed particularly well when forecasting diseases that follow predictable progression patterns, such as certain types of cancer.
• The model outperformed existing machine-learning algorithms that use biomarkers—levels of specific molecules in the body—to predict multi-disease risk.
• When tested on 1.9 million people in the Danish National Patient Registry, predictions were only slightly less accurate than UK Biobank results.

Why this matters: Current healthcare approaches require running dozens of separate risk assessment tools to get comprehensive disease predictions.

• “A health-care professional would have to run dozens of them to deliver a comprehensive answer,” says study co-author Moritz Gerstung, a data scientist at the German Cancer Research Center in Heidelberg.
• The tool’s multi-disease modeling capability could revolutionize preventive care by identifying patients who need early intervention across multiple health conditions.

What experts are saying: The research community is impressed by the model’s comprehensive approach to health prediction.

• “It can generate entire future health trajectories,” says Stefan Feuerriegel, a computer scientist at the Ludwig Maximilian University of Munich who develops AI models for medical applications, calling the multi-disease modeling “astonishing.”
• “It worked astonishingly well,” Gerstung noted about the model’s performance during testing.

Current limitations: The model faces several constraints that researchers are working to address.

• UK Biobank data only captured participants’ first occurrence of each disease, missing important information about recurring conditions.
• “The number of times someone has had an illness is important for the modelling of personal health trajectories,” says Degui Zhi, a bioinformatics researcher at the University of Texas Health Science Center at Houston.
• The model was trained exclusively on UK data, potentially limiting its accuracy when applied to other populations.

What’s next: Researchers plan to expand Delphi-2M’s scope by testing it on datasets from multiple countries to improve its global applicability and precision.