Scientists at Imperial College London developed CardioKG, an AI tool that identifies genes linked to heart disease and accelerates drug discovery by combining heart imaging data from the UK Biobank with large medical databases. Cardiovascular diseases cause 1.7 million deaths annually in the European Union.
Cardiovascular diseases rank as the leading cause of death and disability across the European Union, according to the Organisation for Economic Co-operation and Development. These conditions affect 62 million people in the region. The new study introduces CardioKG as a response to this burden, enabling more precise identification of potential treatments.
CardioKG relies on heart imaging data collected from thousands of participants in the UK Biobank. This dataset encompasses patients diagnosed with atrial fibrillation, heart failure, and heart attacks, alongside healthy volunteers. Researchers utilized these detailed scans to link structural heart information with genetic and pharmacological data.
The tool employs knowledge graphs, which connect disparate elements of medical information. “One of the advantages of knowledge graphs is that they integrate information about genes, drugs and diseases,” said Declan O’Regan, group leader of the Computational Cardiac Imaging Group at the MRC Laboratory of Medical Sciences, Imperial College London. This integration allows for predictions on which medicines could address specific heart conditions with greater accuracy.
Researchers indicate that the methodology supports personalized care by aligning treatments with an individual’s heart function patterns derived from imaging. The approach extends beyond heart disease, as the imaging-based knowledge graph can adapt to other medical imaging applications, such as those for brain disorders and obesity.
Incorporating heart imaging data into the knowledge graph enhanced the detection of novel genes and drugs. “This means you have more power to make discoveries about new therapies. We found that including heart imaging in the graph transformed how well new genes and drugs could be identified,” said O’Regan.
The analysis identified methotrexate, a drug commonly prescribed for rheumatoid arthritis, as a candidate for treating heart failure patients. Gliptins, a class of medications used for diabetes management, emerged as potentially beneficial for individuals with atrial fibrillation.
Additional findings pointed to a possible protective effect of caffeine in certain patients with atrial fibrillation. Researchers emphasized that this observation does not justify alterations to caffeine consumption habits.
Future developments will evolve CardioKG into a dynamic, patient-centered framework. “Building on this work, we will extend the knowledge graph into a dynamic, patient‑centred framework that captures real disease trajectories,” said Khaled Rjoob, the first author of the study and a data science researcher at Imperial College London. This extension aims to facilitate personalized treatment options and predictions of disease onset.
