As artificial intelligence and machine learning continue to evolve, they're starting to have real-world impacts on how physicians do their jobs. But some docs are also skeptical if not outright scared that AI might be coming for their jobs.

That's the wrong way to look at it, said cardiologist Anthony Chang, MD, chief intelligence and innovation officer at Children's Hospital of Orange County, who's excited about what he says is an "amazing paradigm shift in medicine."

It can perhaps be equally exciting and disconcerting seeing IBM's Watson winning on Jeopardy! and AI technology from Google's DeepMind subsidiary edging out Ke Jie, the world's best (human) player of the ancient Chinese strategy game Go.

With computerized smarts like that, it's no wonder that many physicians, particularly radiologists, harbor concerns that their hard-earned wisdom could eventually be replaced by artificial intelligence.

But Chang, who's also the founder of AIMed: Artificial Intelligence in Medicine, takes a much more optimistic view of what AI and machine learning can do for healthcare.

"It's a very common misconception that AI in its various forms will displace the clinician from certain subspecialties," he told HIMSS Innovator-in-Residence Adam Culbertson at the recent Dev4Health 2018 Conference. "But I would look at it the other way around: How do we incorporate machine learning and other aspects of AI to augment the intelligence of the clinician?"

A report earlier this year on machine learning in medical imaging found that a sizable majority of those polled – directors of radiology, imaging directors, radiology managers, PACS admins and others – said AI is either "important or extremely important” in the field. Just 16 percent said they don't think it will have much of an impact.

It will, said Chang, and it will be a positive one.

In his formulation, "machine intelligence," when combined with "human intelligence," leads to what he calls "medical intelligence."

That synergy is important, he said. "Both man and machine can learn from each other for the better."

Radiologists shouldn't freak out that machines are reading medical images, in other words. They should by thinking creatively about how machine learning can decrease the burden on the radiologist.

Imaging data is "exponentially growing in volume," he noted. Soon there's not going to be enough doctors to read it all. That's especially true in the developing world, Culbertson pointed out, which could offer an intriguing use case for the application of AI.

"As a cardiologist, I read a lot of normal echoes," said Chang. "I would rather my time be spend reading echoes that are equivocal or need a high level of interpretation."

Artificial intelligence can "make the job more interesting for doctors," he said – interpreting data on standard test results and freeing up more time for the humans to focus on the "interesting cases, and not the normal cases."

Twitter: @MikeMiliardHITN
Email the writer: mike.miliard@himssmedia.com