Scientists have unveiled an artificial intelligence system capable of predicting an individual’s risk of developing health problems years in advance. The model, known as Delphi-2M, can assess the likelihood of more than 1,200 conditions by analysing patterns in medical records, much like a weather forecast gives probabilities of rain. Researchers believe this could transform prevention and healthcare planning by identifying high-risk patients and forecasting future hospital demand.

Unlike predicting the exact timing of an event, Delphi-2M calculates the probability of diseases such as type 2 diabetes, heart attacks, and sepsis. The tool was trained using anonymous medical data from over 400,000 participants in the UK Biobank project and later tested with 1.9 million Danish records. Results suggest the system is accurate, with predictions closely matching real-world outcomes.

The researchers emphasise the model is not yet ready for clinical use but could one day guide interventions, from prescribing preventative medicines to giving tailored lifestyle advice. For example, people at higher risk of liver conditions could benefit more from reducing alcohol than the general population. It could also help policymakers anticipate healthcare needs in specific regions, such as estimating the number of heart attacks in a city years ahead.

Published in Nature, the study highlights both the promise and limitations of the technology. The dataset used mainly covered adults aged 40 to 70, raising concerns over bias, while further refinement is needed with genetic, imaging and blood data. Still, experts say Delphi-2M marks the beginning of a new era of predictive healthcare that could eventually personalise treatment and improve resource planning.