Assimilated LVEF: A Bayesian technique combining human intuition with machine measurement for sharper estimates of left ventricular ejection fraction and stronger association with outcomes

2018/05/07 by McAndrew, Thomas, Redfors, Bjorn, Crowley, Aaron +11
#Applications (stat.AP) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.1805.02764

Abstract

The cardiologist's main tool for measuring systolic heart failure is left ventricular ejection fraction (LVEF). Trained cardiologist's report both a visual and machine-guided measurement of LVEF, but only use this machine-guided measurement in analysis. We use a Bayesian technique to combine visual and machine-guided estimates from the PARTNER-IIA Trial, a cohort of patients with aortic stenosis at moderate risk treated with bioprosthetic aortic valves, and find our combined estimate reduces measurement errors and improves the association between LVEF and a 1-year composite endpoint.

Related

• Development of Automated Neural Network Prediction for Echocardiographic Left ventricular Ejection Fraction
• EchoNarrator: Generating natural text explanations for ejection fraction predictions
• Lightweight and Interpretable Left Ventricular Ejection Fraction Estimation using Mobile U-Net
• Anatomically Faithful but Temporally Diffuse: Auditing Attribution for Left-Ventricular Ejection-Fraction Estimation from Echocardiography
• Cyclical Self-Supervision for Semi-Supervised Ejection Fraction Prediction from Echocardiogram Videos