Grant: Project Grant

Identification of patients at increased risk for stroke is most often based on clinical factors including the age of the patient and the presence or absence of conditions such as abnormal heart rhythm, hypertension, diabetes, and heart failure. Risk assessments that rely only on these clinical factors have many limitations. They do not, for example, take into account differences in stroke risk between patients with well controlled hypertension versus those with uncontrolled blood pressure. Previous research has shown that changes in the structure and function of the heart can influence risk for stroke. Echocardiography is a non-invasive imaging technique that is widely used to rapidly obtain information on heart structure and function. We propose to use artificial intelligence to analyze large amounts of echocardiography image data to identify features associated with increased risk for stroke. Our research group has previously developed machine learning approaches to automate echocardiography image analysis and will apply this technology to clinical and echocardiography data collected at large tertiary care centres over the past 10 years to find new ways to predict disease in patients at risk of stroke. The new imaging risk factors identified in this study will then be tested in 10,000 patients to confirm their performance. We expect that the risk factors identified in this study can be combined with clinical assessment to improve patient care by identifying more individuals at risk of stroke, earlier in their disease, so they can be started on effective preventative treatments sooner.