How to bridge the imaging AI adoption gap with an ecosystem approach — 3 learnings

Imaging technologies powered by artificial intelligence have proliferated in recent years, yet adoption of these technologies, which are designed to assist with the complexities of diagnostics and their associated workflows, has not kept pace.

A host of barriers and the nature of rapidly evolving technology are part of the reason. It raises the question, what's the best way to bridge the gap?

In a webinar hosted by Becker's Hospital Review and sponsored by IBM Watson Health, two physicians who are imaging experts discussed challenges facing AI adoption in the imaging world, and ways that technology vendors and providers can overcome these barriers.

  • David Gruen, MD, chief medical officer for imaging, IBM Watson Health
  • Jonathan Messinger, MD, medical director of diagnostic imaging, Baptist Health South Florida in Miami

Three learnings:

1. The number of imaging AI technologies has exploded, but provider adoption of these technologies has not kept pace. There are many barriers to adoption among providers. Radiologists are concerned that such technology could replace them. In a recent study published in the journal Clinical Imaging, 60 percent of radiology interns view the prevalence of AI as a negative influence on choosing a specialty. Fewer than 30 percent of radiologists are using AI as part of their practice currently, according to research published in the Journal of the American College of Radiology. Additional concerns include design bias, reliability and challenges of finding the right solution in an ever-changing market.

2. Healthcare lacks a holistic ecosystem approach to AI. There are many stakeholders to weigh in on the value, role and use of technology in healthcare. In order to be successful, AI must meld its capabilities in a way that it does not look at just one specific diagnostic problem but looks broadly at all potential problems. For example, stroke evidence is just one facet of an image; a physician also needs to be able to account for other markers of disease, like dementia, multiple sclerosis and white matter disease. With capable AI, this sort of "algorithm orchestration" would allow valuable insights to be rapidly delivered to the radiologist's workstream in a useful way.

"We all seem to have our heads down rather than up about how we work together," Dr. Gruen said. "As an industry, we've looked at this as, 'If we build it, you will come,' instead of, 'If we partner together to create good solutions we'll add value to our patients and our providers.'" Pivoting to an ecosystem approach focused on beneficial outcomes for all is a more effective approach, he said.

3. Measuring success still has to be figured out. AI technology is new and rapidly changing, and many providers aren't yet sure how to go about evaluating the technology against the models they are used to: Get a product with a proven return on investment; reevaluate after five or 10 years. And because it is so new, how does a clinical team — whose members are increasingly tasked with being front-line recommenders and evaluators of new technology — measure success and show return on investment? "How do we tell that it is actually solving the problem that we stated at the onset?" Dr. Messinger asked. "We're still trying to figure those things out."

While barriers to adoption remain, along with questions about proving the technology's value, it's clear that AI in imaging is going to be a key diagnostic tool. How health systems go about adoption and integration will require teamwork and a shared vision.

To register for upcoming webinars, click here.

Copyright © 2024 Becker's Healthcare. All Rights Reserved. Privacy Policy. Cookie Policy. Linking and Reprinting Policy.

 

Featured Whitepapers

Featured Webinars