ѻý

Is Adding an AI Readout to Mammography Worth It?

<ѻý class="mpt-content-deck">— Women would have to pay out-of-pocket, as reimbursement is a long way off
MedpageToday
A photo of a female physician holding a mammogram and filling out a form.

More women may be getting an offer to add an artificial intelligence (AI) readout to their mammograms, but they'll have to pay for it out-of-pocket -- and many are wondering if it would even result in better outcomes.

Indeed, the jury is still out on the evidence behind the technology. But radiologists -- who have been at the forefront of testing AI for clinical uses -- are likely to continue to offer the option, which experts say can run as high as $100 for a single screening.

"It's new technology. There's an expense with that new technology. Who pays for it?" Constance Lehman, MD, PhD, of Massachusetts General Hospital, told ѻý. "That's the very uncomfortable situation we're in right now."

have reported that AI algorithms improve cancer detection. One such company, RadNet, introduced AI-assisted mammography in New York and New Jersey last year and has since expanded to several other states, .

RadNet officials told the news outlet that its own research showed their doctors' performance improved when they used AI to read mammograms. The company did not return a request for comment from ѻý.

Much of the published research on AI and mammography has come from Europe, not the U.S. Interim results from the of 80,000 Swedish women, for instance, found cancer detection rates were 20% higher when AI did an additional read.

But whether these results will be the same in America remains to be seen, and other research has also shown that AI-assisted detection can produce inconsistent outcomes.

While the evidence gets sorted out, reimbursement will remain an issue, experts said. Lehman noted that AI-assisted detection has not been added to the array of codes that streamline coding for such medical services.

There aren't any talks about creating a CPT code for AI-assisted detection, either, she said: "We're not close to that, yet."

Lehman compared the dilemma to what happened when computer-assisted detection (CAD) for mammography was cleared by the FDA for clinical use in 1998. Eventually, CAD was added to CPT lists, and those codes are now used as part of a bundled payment for mammography screens.

She also noted that 3D mammography, or tomosynthesis, didn't initially have a CPT code and wasn't reimbursed by many payers for some time. Now, it's covered by most insurers, she said.

Bibb Allen, MD, the chief medical officer of the American College of Radiology's (ACR) Data Science Institute, said the heart of the problem is that payers will be reluctant to reimburse for the technology until more favorable evidence is published.

"There are lots of reasons that AI reimbursement is sort of a square peg trying to get into the round hole," he said.

Allen believes use of AI alongside a radiologist can provide better care than either one alone, but more research will be needed to prove that definitively, he said. Lehman agreed that the research needs to catch up first.

"We've got decades of science and research and understanding of how these tools ... help a radiologist do better at finding cancer on a mammogram, where they've done well, and where they failed," Lehman said. "We're leveraging that knowledge and bringing it forward in the new era of AI."

  • author['full_name']

    Michael DePeau-Wilson is a reporter on ѻý’s enterprise & investigative team. He covers psychiatry, long covid, and infectious diseases, among other relevant U.S. clinical news.