Swiss public broadcaster RSI covered two AI-CVM studies showing that machine learning models can predict the clinical course of cardiac patients more accurately than established risk scores. The report, “Cuore e intelligenza artificiale: nuovi strumenti predittivi”, describes work led at the Department of Cardiology, Inselspital, University of Bern, with Christoph Gräni: a model published in The Lancet Digital Health that improved long-term prognostic accuracy in severe aortic stenosis by 10 to 15% over standard risk models, developed on 2,985 patients in Bern and tested in more than 1,000 patients in Japan, and a model published in JAMA Cardiology that improved prediction of death or heart failure hospitalisation in transthyretin cardiac amyloidosis by up to 20% compared with current international staging systems, based on 850 patients from Swiss and Austrian centres. As the piece notes, these tools analyse routinely collected data such as laboratory values, clinical records and imaging to support individualised risk assessment; they do not replace the cardiologist.

Link to article: https://www.rsi.ch/info/scienza-e-tecnologia/Cuore-e-intelligenza-artificiale-nuovi-strumenti-predittivi–4051665.html