Artificial intelligence is ready for doctors. But doctors might not be ready for it

Welcome back to Path to Recovery, a newsletter that will bring you weekly conversations on how the health care profession will recover from one of the most significant crises of our time. Click "subscribe" above or follow along using #PathtoRecovery.

Can a robot read a CT scan better than your doctor? I posed that question four years ago in an article on LinkedIn. But since then, the debate around artificial intelligence has become less about “can it do what we want it to do” and more about the role it will play in medical practice.?

AI has shown a lot of potential (and, to be fair, plenty of hype). But some of the most pressing questions involve what the technology will mean for today’s practicing physicians.?

While there are a number of applications for AI in healthcare, the field of radiology is at the forefront of many of these debates. So it’s not coincidental that the annual meeting of the Radiological Society of North America remains one of the top conferences for vendors showing off new technologies.

This year was no exception. WIth a theme of “Redefining Radiology,” artificial intelligence featured prominently in this year’s #RSNA21 program, with more than 100 sessions on the topic, a huge increase over four years ago, when there were about two dozen.

There was plenty of excitement to go around.

This year, like last, also included presentations on the promise of AI in detecting and furthering our understanding of COVID-19.?

Yet in addition to clinical discussions, many of the AI sessions focused on practical issues such as how to balance the cost of AI with its promise, how to fit it into current workflows and the importance of making sure datasets include a diverse group of patients. While AI systems excel at pattern recognition, much of the art of medicine is in being able to recognize the outliers.?

That real-world return on investment is perhaps why most radiologists seem to be taking a wait-and-see approach before adopting the technology, despite the fanfare.?

Only 29% of those in the radiology field, when surveyed by medtech company Philips, said they’re currently adopting AI tools, while 82% see an investment in the next three years. That slower approach puts radiology behind cardiology, another field poised to benefit from AI; 40% of cardiology respondents said they’re currently making an investment in the technology.?

One potential reason for the delay, however, might have to do with competing priorities; more than two-third of radiologists in Philips survey indicated that they’ve been focusing on investments in telehealth.

But as far as the original “can it?” question, most doctors agree that they’re not being put out of business just yet. “The human component of medicine is a long way from being automated and replaced by machines,” writes Dr. Eugene Ong , a dermatologist and investor in this space, in an article on LinkedIn . “The outputs of diagnoses from machines will still have to be conveyed to patients with human sensitivity, tact, and emotional understanding. In that, for now, machines cannot compete.”

Still, one of the truisms about AI in healthcare is that the technology won’t replace doctors – but doctors who know how to use AI might replace those who don’t. When that tipping point will occur is still anyone’s guess.?

I’d love to hear your thoughts in the comments below.

For more conversations on this topic, here are five radiology voices to follow:

Dr. Jinel Scott , director of emergency radiology, quality and patient safety at Kings County Hospital

Dr. Randy Miles , assistant professor of radiology at Harvard Medical School

Dr. Geraldine McGinty , senior associate dean for clinical affairs at Weill Cornell Medicine

Dr. Carlo De Cecco , associate professor of radiology and biomedical informatics at Emory University School of Medicine

Prof. Dr. med. Mathias Goyen , chief medical officer, EMEA, at GE Healthcare