Scientists develop AI Model to predict patient health conditions years in advance

An international team of researchers has developed an advanced artificial intelligence model capable of forecasting patients’ medical conditions and potential diseases years before onset.

The model, named Delphi-2M, was introduced in a study published in Nature by experts from institutions in the UK, Denmark, Germany, and Switzerland. Based on patients’ medical histories, Delphi-2M can predict the likelihood of more than 1,000 diseases.

According to the research, the model was trained using data from the UK Biobank—a large-scale biomedical database—employing neural networks and transformer-based systems, the same underlying technology that powers ChatGPT and other language models.

Speaking to international media, Dr. Moritz Gerstung, an AI expert at the German Cancer Research Center, explained that identifying disease progression is similar to learning grammar in a language.

Delphi-2M analyzes patient data to recognize patterns, learning which illnesses emerge in sequence or occur together, thereby enabling highly meaningful and medically relevant predictions.