Obstructive coronary artery disease diagnostics: machine learning approach for an effective preselection of patients

Most patients undergoing diagnostic imaging of coronary artery disease do not show significant coronary artery stenosis. Despite this, they are exposed to the negative effects related to the radiation and administration of the contrast agent. An effective preselection of patients for further diagnostics may be the solution. There are reports of machine learning algorithms adapted to perform such a task. In this chapter we briefly introduce the reader to the problem of obstructive coronary artery disease and do our best to explain how to develop such an algorithm. The work is based on clinical data and diagnostic results collected at Silesian Center for Heart Diseases in Zabrze, Poland.