Why artificial intelligence applied to the electrocardiogram is not yet clinical routine?

Abstract

Artificial intelligence opens up multiple application scenarios to the electrocardiogram (Ai-Ecg) in various clinical settings, with potentially extremely interesting expected results. However, the introduction of prediction models of clinical diagnosis, of the results of different therapies and of the different clinical risks based on artificial intelligence requires that the limits of this approach be adequately considered, limits ranging from conflicts of interest to the transparency of procedures, from product certification to cyber security, to the need for coordination between the most diverse professional figures involved. In order for the Ai-Ecg to become a routine clinical resource, it is necessary to institutionally invest in projects that have as their objectives an ever greater and more extensive attention to its developments and its applications in vast and heterogeneous extended clinical contexts.