An Automated Machine Learning-Based Quantitative Multiparametric Approach for Mitral Regurgitation Severity Grading.

Area of Science:

• Cardiology
• Artificial Intelligence in Medicine
• Medical Imaging Analysis

Background:

• Mitral regurgitation (MR) is common, but its severity assessment is subjective and variable.
• An automated tool can standardize MR grading and improve patient management.

Purpose of the Study:

• To develop and validate a fully automated machine learning (ML) workflow for grading MR severity using echocardiography.
• To assess the accuracy and efficiency of the ML model in classifying MR severity.

Main Methods:

• ML algorithms were trained on echocardiograms from two cohorts and validated on two independent studies.
• The model measured 16 MR-related parameters, with the optimal model using 9 parameters.
• Ground truth was established by a multiparametric echocardiography core laboratory assessment.

Main Results:

• The automated workflow achieved 97% accuracy in identifying significant (moderate or severe) MR.
• Image analysis was rapid (80 ± 5 seconds per case) and feasible in 99.3% of cases.
• Severe MR grading by the model predicted higher 1-year mortality (adjusted HR: 5.20).

Conclusions:

• A multiparametric ML model for automated MR severity grading is feasible, accurate, and fast.
• This tool can streamline diagnosis, facilitate timely referrals, and improve patient outcomes.
• The model's ability to predict mortality highlights its clinical significance.