Artificial intelligence and medicine: The need for interpretable models

ABSTRACT

The pandemic has provided clear examples of the potential of AI for the health sector, as well as some of its issues, largely derived from the use of black box models. In some cases, there are no reasonable alternatives, as in image and speech processing. However, in many other instances it would be more profitable to try to focus the developments on Interpretable AI, which could be used more directly for the confirmation of knowledge or for the generation of new hypotheses that can be tested with subsequent experiments. © Global Knowledge Academics, authors. All rights reserved.