“Most of the medical technologies that use artificial intelligence (AI) are based on conventional ones.”
So said Park Seong-ho, a professor of radiology at Asan Medical Center, while presenting the results of a study how to evaluate AI-based healthcare technology in radiology for insurance coverage.
The Health Insurance Review & Assessment Service (HIRA) commissioned the research and Park unveiled the results at a symposium on “Regulatory Innovation for Innovative Medical Technology and Device” at GS Tower in Yeoksam-dong, southern Seoul, Wednesday.
As developers rush to develop AI-integrated healthcare technologies recently, they are eager to know whether their advanced technologies will be reimbursable.
However, Park made it clear that AI-based medical technologies are facing limitations in receiving health insurance coverage.
“There is no case that AI-based medical technology has been covered by insurance in the world,” Park said. “Obtaining recognition for novel medical technology means receiving a new code for reimbursement. But currently, the government has the same review standards. It’s only different who analyze and read AI images, a radiologist or an AI machine.”
Lee Sang-moo, an assessment committee member at HIRA, said the government considers an AI-based medical technology as an existing treatment if it simply assists a physician or fails to show additional therapeutic improvement or value.
“If a company offers an unprecedented medical technology, we regard it as a new technology. From an insurer’s point of view, technological innovation is not important. We see how much it benefits patients when judging innovation,”he said.
Lee also expressed his anticipation for AI in the healthcare sector, as well as related concerns.
“AI will have many merits such as assisting physicians to reduce misdiagnosis and enhance patient safety, and to help them seeing patients more quickly,” he said.
However, relying too much on AI could oversimplify the problems of complex diseases, undermine quality nurturing of experts who will verify AI-based results, and raises the possibility of errors in a massive scale, he warned.