TLDR: European scientists have developed Delphi-2M, a generative AI model trained on 400,000 medical records from the UK Biobank. This innovative system can predict an individual’s susceptibility to over 1,000 diseases up to 20 years before symptoms appear, offering a new frontier in preventative medicine.
In a significant leap forward for predictive healthcare, European scientists have unveiled ‘Delphi-2M,’ a generative artificial intelligence model capable of forecasting an individual’s risk of developing more than 1,000 diseases decades into the future. The research, published in Nature on September 17, 2025, highlights the model’s potential to revolutionize preventative medicine.
Delphi-2M was rigorously trained on an extensive dataset comprising 400,000 anonymized medical records sourced from the UK Biobank, a comprehensive data repository. This massive training regimen allowed the AI to identify intricate patterns of disease progression that are often imperceptible to the human eye. Consequently, the system can predict future disease risks based on a person’s current medical record, with some predictions extending up to 20 years in advance of symptom onset.
Traditional AI health prediction systems typically focus on estimating the risk for a single disease. In contrast, Delphi-2M’s ability to assess susceptibility across a vast spectrum of over 1,000 conditions simultaneously marks a substantial advancement. Researchers, including Moritz Gerstung and Artem Shmatko, are optimistic that this multi-disease predictive capability could significantly streamline healthcare processes, saving valuable time for medical professionals, and providing crucial insights for calculating disease burdens at a population level.
Also Read:
- Penn Researchers Pioneer Generative AI for Potent Antibiotic Discovery
- Fred Hutch Leaders Secure $10 Million for AI Startup Synthesize Bio to Accelerate Drug Discovery
The implications of Delphi-2M are far-reaching. By identifying individuals at high risk for various conditions long before they manifest, the model could enable earlier interventions, personalized preventative strategies, and more effective management of chronic diseases. This proactive approach could lead to improved patient outcomes and a more efficient healthcare system globally.
