Analyzing breast-cancer tumors with synthetic intelligence has the potential to enhance healthcare effectivity and outcomes. However docs ought to proceed cautiously, as a result of related technological leaps beforehand led to increased charges of false-positive exams and over-treatment.
That is in accordance with a brand new editorial in JAMA Well being Discussion board co-written by Joann G. Elmore, MD, MPH, a researcher on the UCLA Jonsson Complete Most cancers Middle, the Rosalinde and Arthur Gilbert Basis Endowed Chair in Well being Care Supply and professor of drugs on the David Geffen Faculty of Drugs at UCLA.
“With out a extra strong method to the analysis and implementation of AI, given the unabated adoption of emergent expertise in medical follow, we’re failing to be taught from our previous errors in mammography,” the JAMA Well being Discussion board editorial states. The piece, posted on-line Friday, was co-written with Christoph I. Lee, MD, MS, MBA, a professor of radiology on the College of Washington Faculty of Drugs.
A type of “previous errors in mammography,” in accordance with the authors, was adjunct computer-aided detection (CAD) instruments, which grew quickly in recognition within the area of breast most cancers screening beginning greater than twenty years in the past. CAD was authorized by the FDA in 1998, and by 2016 greater than 92% of U.S. imaging services have been utilizing the expertise to interpret mammograms and hunt for tumors. However the proof confirmed CAD didn’t enhance mammography accuracy. “CAD instruments are related to elevated false optimistic charges, resulting in overdiagnosis of ductal carcinoma in situ and pointless diagnostic testing,” the authors wrote. Medicare stopped paying for CAD in 2018, however by then the instruments had racked up greater than $400 million a yr in pointless well being prices.
“The untimely adoption of CAD is a premonitory symptom of the wholehearted embrace of emergent applied sciences prior to completely understanding their impression on affected person outcomes,” Elmore and Lee wrote.
The docs counsel a number of safeguards to place in place to keep away from “repeating previous errors,” together with tying Medicare reimbursement to “improved affected person outcomes, not simply improved technical efficiency in synthetic settings.”
Supplies offered by College of California – Los Angeles Well being Sciences. Be aware: Content material could also be edited for type and size.