The less radiologists know about AI, the more likely they are to believe it may displace them from their clinical pursuits.
However, the inverse is also true—more AI knowledge, less AI fear—so AI should be taught in radiology residency programs and, for experienced radiologists, in CME courses.
That’s according to researchers who surveyed more than 1,000 radiologists around the world and had their findings published March 20 in European Radiology.
Merel Huisman, MD, PhD, of University Medical Center Utrecht in The Netherlands and colleagues distributed the survey through various societies, networks and social media platforms.
After tallying the responses, the team used multivariable logistic regression to gauge independent predictors of fear of replacement vs. positive attitudes toward AI.
They found that rudimentary AI-specific knowledge was associated with such job insecurity and some 38% of the cohort fell into this category.
On the other hand, close to half the field, 501 of 1,047 respondents, indicated they had an intermediate or advanced grasp of AI—and this subgroup tended to hold an “open and proactive” attitude toward AI.
Radiology is widely seen as the medical specialty furthest along with AI, so the profession’s perceptions of the technology may serve as medical AI bellwethers.
“Intermediate and advanced AI-specific knowledge levels may enhance adoption of AI in clinical practice, while rudimentary knowledge levels appear to be inhibitive,” Huisman et al. comment. “AI should be incorporated in radiology training curricula to help facilitate its clinical adoption.”
Most of the responding radiologists, 83%, work in Europe, leaving North America, South America, Asia and Africa underrepresented. Just 6% (n = 64) work in the U.S. or Canada. Consequently, the results “should be interpreted as a reflection of the opinion of the radiology community in western society, mainly Europe,” the authors point out.
The study is available in full for free.