Last year, at the Machine Learning Conference in
Toronto, Geoff Hinton, Ph.D., one of
world’s leading experts on artificial intelligence, remarked that “if you work
as a radiologist, you are like the coyote who is already over the cliff but
hasn’t yet looked down and so doesn’t realize that there is no ground
underneath him. I think we should stop training radiologists now.”
The reason Dr. Hinton offered this gloomy prediction was
that artificial intelligence has the potential to someday become even better
than humans at reading medical images. When this day comes, it will create a
sea change for radiologists and the business of radiology practice management.
On the third day of the RBMA’s Paradigm conference in Chicago,
attendees turned their attention toward this issue with AI. They started the
day with a presentation from Christopher Austin, M.D., Global Radiology
Solutions Director at GE Healthcare, and ended it with a presentation by Keith
Dreyer, D.O., Ph.D., FACR, FSIIM, one of the leading experts on computer
applications to medical imaging and chair of the American College of
Radiology’s IT and Informatics Committee.
Both Dr. Austin and Dr. Dreyer agreed that AI would require
a major paradigm shift in how radiologists and radiology practices view their
role in patient care. But they disagreed
with Dr. Hinton on the potential for AI to put radiologists and radiology
practices out of business.
Rather, AI would allow radiologists to focus their energies
on the jobs AI cannot do as well, such as those that require human creativity.
The image Dr. Dreyer evoked was of the centaur, the mythical creature that was
half man and half animal. In the future, radiologists will be able blend their
human evolutionary intelligence with artificial intelligence tools to create
something greater than either type of intelligence can offer on its own.
Furthermore, the RBMA and its members will need to be there
to guide the implementation of AI and work out all those pesky business details
the visionaries pass over, like will AI reads be part of the professional or
technical component of imaging? And what about liability? How many RVUs for an
AI read compared to a human one? And how do you get FDA approval on a computer
tool that is continuously learning and changing?
Issues covered in the educational sessions between Dr.
Austin and Dr. Dreyer’s presentations only underscored the breadth of topics
the RBMA will need to continue to cover, including marketing, leadership
development, regulatory concerns and payment reform. Plus, association members
were heavily engaged in something AI will never be able to do, coming up with
new creative ideas through networking and friendships with other intelligent
members of their field.
As new RBMA executive director Bob Still noted at the
Paradigm conference’s networking lunch and awards ceremony, there has never
been a better time than now to join the RBMA.