Recent advancements in artificial intelligence (AI) have led to the development of a model capable of predicting individual health issues up to a decade in advance. Researchers have created this AI, named Delphi-2M, to analyze patterns in anonymous medical records, enabling it to estimate the likelihood of over 1,000 diseases. This predictive capability draws a parallel to weather forecasting, where probabilities, such as a 70% chance of rain, are communicated to the public.
Delphi-2M utilizes a methodology akin to that of AI chatbots, which are trained to recognize patterns in language. The model has been developed using a substantial dataset from the UK Biobank, encompassing medical history and lifestyle factors from over 400,000 individuals. Validated against medical records of 1.9 million people in Denmark, initial findings indicate that its predictions align closely with actual outcomes, particularly for diseases with clear progression, such as type 2 diabetes and heart attacks.
The primary goal of this research is to identify high-risk patients early, allowing for preemptive interventions that could include tailored lifestyle advice or medications. Although the model has not yet reached clinical application, its potential extends to informing disease-screening programs and aiding healthcare resource planning.
However, the model is still in the research phase and requires further testing and refinement to address potential biases, particularly since its initial data is predominantly from individuals aged 40 to 70. Future updates aim to incorporate additional medical data types, including genetic and imaging information. Prof. Ewan Birney, one of the lead researchers, emphasizes the importance of thorough evaluation and regulation before clinical use, likening its development timeline to that of genomic medicine.