A novel, fully automated, quantitative, multiparametric machine learning (ML) model for grading the severity of mitral regurgitation (MR) is feasible and highly accurate, according to new study results. Federico M. Asch, MD, of the MedStar Health Research Institute at MedStar Washington Hospital Center and Georgetown University, reported these results Thursday during a late-breaking trial presentation at the European Association of Cardiovascular Imaging annual meeting in Barcelona, Spain. The most common vascular heart disease in the U.S., and the second most common in Europe, is MR. As the severity of MR increases, its association with high morbidity and mortality also increases. Grading MR is a difficult task, and this study examined how ML and artificial intelligence (AI)-based models can improve the process of MR detection and therapeutic pathways. This study sought to formulate and validate an AI-based model to grade severe MR. The first phase trained and tested the model for measurements and image analysis using 1,190 patients from two cohort studies split into independent, random groups. The second phase measured the development of ML models for MR severity using 438 patients from two large clinical trials – COAPT and PROMIS-HFpEF. Three MR Doppler parameters (vena contracta, regurgitant jet area ratio and continuous-wave Doppler density) were measured in several cardiac cycles, frames and views. The echocardiographic core laboratory at MedStar Health graded the severity of MR in patients using to the following classifications: none/trace, mild, moderate or severe. The multiparametric model was feasible in 99.3% of cases, and ML variability was zero. The majority voting method had the best sensitivity as well as the best accuracy. Thus, the fully automated, quantitative multiparametric ML model works for grading MR severity because it is fast, usable and extremely accurate. “Its implementation in clinical practice could improve patient care by facilitating proper referral to specialized clinics and increasing access to evidence-based therapies while improving quality and efficiency in the echocardiography laboratory,” Asch concluded.