Two international studies led by the AI in Cardiovascular Medicine (AI-CVM) research group at the Department of Cardiology, Inselspital, Bern University Hospital, and the University of Bern have been highlighted in the latest UNIBE and Inselspital news.

Both studies show that artificial intelligence can predict the course of heart disease more accurately than today’s established risk models. The first developed a multimodal AI model to predict long-term mortality after transcatheter aortic valve implantation (TAVI), trained on nearly 3,000 patients from Bern and externally validated in over 1,000 patients from Japan, improving prognostic accuracy by roughly 10–15% over established risk scores. The second study developed a machine learning model for transthyretin amyloid cardiomyopathy that predicts death or heart failure hospitalization, based on data from 850 patients across Swiss centers and the Medical University of Vienna. It outperformed the established Mayo and NAC scores by several percentage points, up to nearly 20%.

“Two people with the same heart disease often carry a very different risk of complications. Our models help us detect these differences more reliably and tailor follow-up and treatment to the individual patient. AI supports clinical decision-making, but does not replace it,” says Prof. Christoph Gräni, Senior Physician at the Department of Cardiology and senior author of both studies.

“These results show how much potential lies in close collaboration between medicine, computer science and data science. Only together can we develop AI that delivers real benefit for patients,” says Dr. Isaac Shiri, Head of the AI in Cardiovascular Medicine research group.

Both models are freely available online for research use.