Predicting Cardiovascular Events Using Machine Learning

Diseases of the heart and vessels are the most common cause of death worldwide and often enormously limit patients quality of life. Atherosclerosis is the most common pathological change to the arteries, and it is characterized by chronic progression as well as hardening, thickening, loss in elasticity and narrowing of the blood vessels. Since atherosclerosis does not cause symptoms for a long time, it often remains undiscovered and can cause fatal cardiovascular events such as heart attacks or strokes. Early identification of people at high risk of developing atherosclerosis is thus very important so that preventive measures can be taken. The international research project PRE-CARE ML aims to use the large amount of available yet unexploited health data to predict cardiovascular events. As digitization increases, this data is becoming more extensive, but it cannot be sufficiently analyzed and exploited using conventional methods. The remedy for this may be risk forecasting tools based on artificial intelligence, which should be enhanced and made more widely available. Scientists from the Medical University of Graz and the Styrian State Hospital network (KAGes) have teamed up with scientists from Karolinska Institutet in Stockholm, the Hasso Plattner Institute in Potsdam, the Icahn School of Medicine at Mount Sinai, New York, the University of Sao Paolo, Brazil, and the University of Maribor, Slovenia and established an international research consortium with the aim to predict and prevent fatal cardiovascular events. Medical information is becoming increasingly digitized, which means gigantic amounts of electronic health data are available for risk prediction. Standard approaches fail to process this data completely and make it available for medical problems and forecasting. PRE-CARE ML aims to use artificial intelligence methods to develop modern risk forecasting tools for early detection of people at a high risk for cardiovascular disease. Scientists are falling back on their prior experience in using machine learning algorithms for risk prediction in order to enhance them as part of a multidisciplinary consortium and to validate and enhance models of different hospital networks and population groups. PRE-CARE ML works closely with hospitals to integrate models into their information systems and evaluate the impact on daily hospital routines. The researchers will also look into effective communication strategies in order to change patient behavior and especially to ensure acceptance of a risk prediction tool by treating physicians at hospitals and in private practices.