Now, what concerns all of you about AI and breast imaging?

So I think one of the things that remains challenging is that there isn't really comparison to priors in the conventional sense of the word.

it only will look at the prior mammograms if it thinks the current mammogram has something suspicious.

And the only effect that the prior mammogram can have is to basically increase or perhaps slightly decrease the probability of malignancy for a finding that it already saw on the current exam and thought was concerning.

And while to some extent, I think all of us read that way at some level, we also try to take a step back for gestalt.

Is there a new asymmetry? Is there a new circumscribed mass? And there are, of course, some cancers, notably invasive lobular cancer that's manifest as a new asymmetry, not infrequently.

And some triple negative cancers can appear to be relatively circumscribed and almost look like a cyst, even on ultrasound.

What do you think about the role of AI in possibly facilitating more personalized risk assessment for breast cancer? Kind of what are your thoughts on the research in this area? And could this also possibly change the role of breast radiologists in the future? I know that's a lot of questions there, so I apologize for that.