Scientists from the Medical Science Laboratory of the Medical Research Council in the UK have developed an artificial intelligence (AI) tool called CardioKG, which generates a detailed view of the structure and function of the heart, identifies previously unknown genes related to heart disease, and predicts two potential therapeutic drugs. This achievement will help accelerate the identification of disease genes, improve the accuracy of drug efficacy prediction, and bring new tools for the diagnosis and treatment of heart disease. The relevant results were published in the latest issue of the journal Nature. The team integrated cardiac imaging data from 4280 patients with atrial fibrillation, heart failure, or heart attack from the UK Biobank, as well as data from 5304 healthy participants. By extracting and using over 200000 image-based features to train the model, and combining information from 18 different biological databases, the CardioKG system learned the diverse changes in cardiac structure and function. The team successfully predicted the association between genes and diseases using CardioKG and explored possible pathways for drug reuse. The test results showed that the model not only identified multiple previously unknown genes related to heart disease, but also predicted two potential drugs for treating heart disease: methotrexate, a rheumatoid arthritis drug, which may improve heart failure; Gliptin, a diabetes drug, may be beneficial to patients with atrial fibrillation. The study also found that caffeine can enhance cardiac excitability and has a certain protective effect on patients with atrial fibrillation accompanied by rapid and irregular pulses. The team believes that the application prospects of CardioKG are not limited to the field of the heart, and similar methods can also be extended to the study of other organs and tissues such as the brain and body fat, providing strong support for exploring new therapies for diseases such as dementia and obesity. (New Society)