A radiologist wrote a piece for Forbes.com earlier this year that had the medical-imaging community buzzing for weeks. Jason Kelly, MD, who practices in Denver, predicted the eventual demise of his profession at the hands of IBM’s Watson and its artificial-intelligence kin.
Once big data equips AI with enough images to learn from, the technology will be able to “tell a hemangioma (benign) from a cholangiocarcinoma (very bad)”—at which point the days of his “caffeine-driven, carbon-based perception machine” will be over, Kelly fretted only half-jokingly.
It was good that he jump-started a jittery national conversation that had been percolating in nervous pockets for months.
Will AI really make radiologists dispensable? (No.) Does it have the potential to all but take over medicine? (No.) Might it make patients in primary care feel like thingamajigs on assembly lines? (They already feel that way—but healthcare economics is more responsible than technology.)
Don’t take it from me.
“I don’t see massive changes, because I think the hype [over AI] is outpacing the reality,” NIH senior imaging investigator Ronald Summers, MD, PhD, told me a few weeks ago. “But I definitely think there is great potential in these artificial intelligence technologies to make a positive impact on patient care.”
Meanwhile, large radiology practices that are already combining deep learning technology with massive clinical datasets are showing how AI “can be used to complement—not supplant—clinicians and improve care,” says Benjamin Strong, MD, chief medical officer of vRad, the tech-intensive teleradiology practice with 400-plus radiologists.
Then again, children in Britain have been observed interacting more readily with Watson than with doctors or nurses, the Medical Futurist website reported in late July.
Children are the future. Watson is getting better, faster and smarter. And it’s homing in on medical imaging in a big way.
Could it be a little soon for the Dr. Jason Kellys of the world to put away their worry stones?