The model was tested on data from 400,000 participants in the UK Biobank—a large biomedical database containing genetic, lifestyle, and medical information about residents of the United Kingdom—as well as data from nearly 2 million people from the Danish National Patient Registry.
What Researchers Are Saying
Professor Thomas Fitzgerald explained, “Medical events have certain patterns, and our model learns to recognize them to predict health outcomes.”
Professor Euan Birney added that in the future, doctors will be able to use the system during consultations: “You will receive a list of major risks and specific advice on how to reduce those dangers.”
Potential Benefits
Delphi-2M is most accurate in predicting diseases that develop gradually, such as or cardiovascular diseases. The authors hope that this technology will help in planning medical resources—such as predicting the number of potential heart attacks in a specific city. Additionally, the model could be enhanced by incorporating biomarkers, imaging results, or even genetic data.
Cautions Exist
Not all experts share this optimism. Professor Justin Stebbing from Anglia Ruskin University warns, “Delphi-2M reproduces biases embedded in the original data.”
Meanwhile, Professor Peter Bannister from the Institute of Engineering and Technology emphasizes, “Before discussing benefits for everyone, we need to ensure digital infrastructure and access to technology for people of all ages and social statuses.”
Why This Matters
The predictions are particularly alarming: according to the One Cancer Voice coalition, by 2040, someone in the UK will be diagnosed with every two minutes. The most common types will remain breast, prostate, and lung cancer, with over 63,000 cases expected among children and young people.
Photo: Unsplash