New AI tool can predict risk of more than a thousand diseases, 20 years ahead
The tool examines a patient’s medical history, lifestyle factors, age, and sex to generate predictions extending up to 20 years
What if an artificial intelligence tool could tell you which diseases you are likely to develop over the next ten or even twenty years?
This is the promise of Delphi-2M, a new AI tool created by scientists from the European Molecular Biology Laboratory, the German Cancer Research Centre, and the University of Copenhagen.
Delphi-2M can forecast a person's risk of more than 1,000 diseases, from cancer and diabetes to respiratory and autoimmune conditions. Unlike traditional tools that focus on single illnesses, it examines a patient's medical history, lifestyle factors, age, and sex to generate predictions extending up to 20 years.
The tool was trained on anonymised records of 400,000 people from the UK Biobank and 1.9 million from the Danish National Patient Registry, making it one of the most ambitious applications of AI in healthcare to date.
"The model learns patterns in medical events and anticipates future health outcomes," said Tomas Fitzgerald, a staff scientist at The European Bioinformatics Institute. Early tests show that its predictions often match or exceed existing single-disease models, and in some cases even outperform biomarker-based algorithms.
However, experts caution that Delphi-2M has limitations. The UK Biobank data primarily captures participants' first encounters with diseases, leaving questions about repeat illnesses unanswered. Moreover, predictions are only as reliable as the underlying data. "It is an intriguing step, but we must be cautious about overreliance on AI for personal health decisions," notes Degui Zhi, a bioinformatics researcher.
Despite the caveats, Delphi-2M hints at a future where doctors might personalise care with unprecedented precision, identifying high-risk patients early and tailoring preventive strategies. As Prof Moritz Gerstung put it, the tool is "the beginning of a new way to understand human health." Whether this future becomes routine clinical practice remains uncertain, but the possibilities are remarkable.
