Artificial intelligence in cardiovascular imaging—principles, expectations, and limitations

Abstract

Seven lessons for cardiologists in the era of artificial intelligence. When reviewing an artificial intelligence application, one needs to critically appraise the following domains: (1) Are artificial intelligence and machine learning necessary to address question of interest? (2) Are the input data of high quality? (3) Is an interpretation of the final algorithm attempted? (4) How is bias addressed? (5) Is appropriate external validation performed? (6) Are there any regulatory or ethical compliance concerns? (7) How much supervision does the final model require?

Topics & Concepts