AI to identify risk of embolism

J. Alberto San Román

Hospital Clínico Universitario de Valladolid, Spain

Infective endocarditis is a rare disease with an annual incidence in Spain of 3.5 cases per 100,000 inhabitants. It is an extremely serious heart disease that leads to death in up to four out of ten patients. In 80 % of cases, mortality is associated with embolic events linked to the disease, such as acute ischaemic stroke. Previous studies have determined that these events are caused by the development of lesions in the inner tissue of the heart.

When this occurs, it is crucial to discern which is the most appropriate option in each case to improve the patient’s prognosis: open-heart surgery to treat the lesion, with the risks that such an intervention entails, or a more conservative treatment. Currently, one of the main factors on which this decision is based is the size of the lesion, although this measure has limited predictive power. As a result, high-risk surgery is sometimes performed on patients who may not need it, while other patients receiving conservative treatment, who might have been saved by timely surgery, develop complications and die.

The aim of this project is to develop a tool capable of efficiently predicting the patient’s prognosis in order to be able to choose the most appropriate treatment at an early stage. To achieve this, artificial intelligence techniques and machine learning algorithms will be applied to analyse images obtained by echocardiography.

Artificial Intelligence system for predicting embolisms in infective endocarditis

Stage 1