Machine learning for assessing the pretest probability of obstructive and non-obstructive coronary artery disease

Abstract

The review presents an analysis of publications on use of machine learning (ML) to assess the pretest probability of obstructive and non-obstructive coronary artery disease (CAD). Data on the high prevalence of non-obstructive CAD among patients referred for coronary angiography are presented, which served as a reason for the development of ML-based models for pretest assessment of coronary anatomy. The use of modern modeling technologies has great potential in verification of obstructive and non-obstructive CAD. It is emphasized that the improvement of prognostic models and their practical implementation is an important element of medical decision making and should be carried out with interdisciplinary cooperation of clinicians and information technology specialists.