Heart disease continues to be one of the leading causes of death worldwide, yet developing new medicines is frequently long, expensive, and unpredictable. A new study approach suggests that answers may already exist—hidden within the human heart and accessible by artificial intelligence. 

Scientists have traditionally used "knowledge graphs" to integrate what they know about genes, diseases, and drugs. A knowledge graph is essentially a large digital map that depicts how various pieces of medical information are related. Until now, these maps lacked one critical component: a realistic representation of how a damaged organ appears and functions in real humans. 

That gap has now been closed with the introduction of CardioKG, a knowledge graph that includes cardiac imaging for the first time. By including comprehensive scans of the heart, the model reflects how the heart's size, shape, and pumping ability differ between healthy persons and those with illnesses such as atrial fibrillation, heart failure, and heart attack. 

To develop this approach, researchers examined cardiac scans from over 9,500 UK Biobank members, including both sick and healthy volunteers. Over 200,000 image-based characteristics were created and integrated with data from 18 biological sources. Artificial intelligence was then utilised to identify relationships between genes, diseases, and existing medications. 

"One of the advantages of knowledge graphs is that they integrate information about genes, drugs, and diseases," the study's authors write. "This means you have more power to make discoveries about new therapies." 

The findings were remarkable. By including real-world heart images, the model improved its ability to predict which genes are involved in heart disease. It also identified unanticipated chances to reuse existing drugs, a practice known as drug repurposing. The algorithm showed that methotrexate, which is often taken for rheumatoid arthritis, could aid with heart failure, and gliptins, which are used to treat diabetes, could benefit individuals with atrial fibrillation. 

Perhaps the most startling finding concerned caffeine. Although caffeine increases heart rate, the model revealed it may have a preventive impact in persons with atrial fibrillation who have rapid and irregular heartbeats. "What's exciting is there are other recent studies in the field which support our preliminary findings," the study's authors say, advocating cautious optimism rather than immediate clinical application. 

Beyond heart disease, the consequences are broad. The same approach might be used in any medical imaging situation, from brain scans in dementia to body-fat imaging in obesity. This technique has the potential to accelerate the development of safer, more tailored medicines by immediately identifying relevant genes and pharmacological targets. 

Looking ahead, the researchers want to create patient-centred models that assess illness progress over time. "This will open new possibilities for personalised treatment and predicting when diseases are likely to develop," the study's authors write.