You should check out the following paper on the need for interpretability in breast cancer research https://arxiv.org/abs/2003.00898

It is a worthy addition to your list of references and is a direct reply to a Google paper on "super human" cancer detection that is completely black-box and had no source/data released for independent testing.