Currently, sonographer technicians provide preliminary interpretations prior to validation and overreading by cardiologists. This staggered, stepwise evaluation allows for the introduction of AI decision support with minimal impact on patient care. Physicians are already used to adjusting the preliminary report given the variable training of sonographers and on the lookout for changes, variation, or adjustments that need to be made.

In preliminary work, a novel AI algorithm developed to assess LVEF was shown to be more precise than human interpretation in 10,030 echocardiograms done at Stanford University (Ouyang et al. Nature, 2020). With randomization, a proportion of the preliminary interpretations will be done by AI technology and the study team will assess how different this preliminary interpretation is from the final interpretation.

A semantic segmentation deep learning model will identify the left ventricle and label the left ventricle. The AI model will produce an assessment of LVEF using video based features.

Proportion of studies the LVEF is changed more than 5% in final report

Mean change in LVEF between preliminary and final report

Proportion of studies the annotation is changed

Mean change in LVEF between prior clinical report and final report