Delphi-2M is a new unique artificial intelligence-based system introduced by European scientists. It is capable of predicting the development of over 1200 diseases – from diabetes to various types of cancer – years in advance. The research results are published in the journal Nature, reported by NOS, writes UNN.
Details
The Delphi-2M system operates on a similar principle to language models like ChatGPT, but instead of texts, it analyzes medical data. The algorithm studies a patient's medical history – diagnosis dates, age, gender, harmful habits such as smoking, or the presence of obesity – and based on this analysis, the system predicts diseases and the possibility of various health problems several years in advance.
Study refutes link between daylight saving time and heart attack risk15.09.25, 17:50 • [views_4302]
The main and key difference of the new system from other similar ones is that it covers almost all known diseases, whereas most similar tools are limited to one specific condition. The system was trained on anonymized data from 400,000 patients from the UK, and its reliability was verified using medical records of almost 2 million Danish citizens.
According to the researchers, the accuracy of Delphi-2M is comparable to existing highly specialized models. The average score is 0.76 out of a maximum of 1.0, and for predictions over 10 years, it is about 0.7.
In some cases, diseases can be predicted even 20 years in advance. The system works best for conditions with a more predictable course – for example, heart attacks or oncological diseases
While doctors' advice based on such predictions may not always completely eliminate the risk, in many cases, patients can reduce it through specific lifestyle changes.
Weight loss or quitting smoking can make a huge difference in preventing certain diseases
This technology can help not only individual patients but also entire healthcare systems. It can predict future healthcare needs, identify priority areas, and plan resources more effectively.
Researchers expect the system to be ready for practical application within a few years. Currently, the development team plans further testing of Delphi-2M in various countries and among patients with different diagnoses.