Release date
New artificial intelligence tools could accelerate the search for treatments for heart disease, a new study has found.
Cardiovascular disease (CVD) is the leading cause of death and disability across the European UnionAccording to the Organization for Economic Co-operation and Development (OECD), approximately 1.7 million people die and 62 million people are affected annually.
Scientists at Imperial College London have developed an artificial intelligence (AI) tool that combines detailed heart scans with large medical databases to identify which genes are associated with disease and help find treatments for heart disease faster.
The tool, named CardioKG, was built using cardiac imaging data from thousands of people in the UK Biobank. This included healthy volunteers as well as patients with diseases such as atrial fibrillation, heart failure, and heart attacks.
By doing this, researchers say they can better predict which drugs will help people with certain heart conditions.
“One of the benefits of the knowledge graph is that it brings together information about genes, drugs and diseases,” said Declan O'Regan, group leader of the Computational Cardiac Imaging Group at the MRC Institute of Medical Sciences at Imperial College London.
Researchers say this approach could ultimately lead to more personalized care, where treatment is better tailored to an individual's heart function.
The same technology could also be applied to studying other conditions using medical images, such as brain diseases and obesity.
“This means we have even more power to discover new treatments. We found that including cardiac images in the graphs changes how accurately we can identify new genes and drugs,” O'Regan said.
Among the drugs highlighted were methotrexate, which is widely used to treat rheumatoid arthritis, and a group of diabetes drugs known as gliptins.
The AI model suggested that methotrexate may be effective in patients with heart failure, while gliptin may be effective in patients with atrial fibrillation.
The analysis also pointed to a possible protective effect of caffeine in some patients with atrial fibrillation, but the researchers stressed that this does not mean caffeine intake should be changed.
“Building on this work, we extend the knowledge graph into a dynamic, patient-centric framework that captures real-world disease trajectories,” said Khaled Lueb, lead author of the study and a data science researcher at Imperial College London.
“This opens new possibilities for personalized treatment and predicting the timing of disease onset.”
