Each week, OMRF Chief Medical Officer Dr. Judith James opens “Adam’s Journal” to answer a medical question from Adam Cohen, OMRF’s senior vice president & general counsel.
Pretty much everywhere I turn, I hear a new story about artificial intelligence. Has AI yet begun to impact the practice of medicine? And how might it affect how physicians and other providers care for patients in the future?
Dr. James Prescribes
As far back as 2020, 30% of radiologists already reported using AI in their practice. The majority said they utilized AI to enhance their ability to read various scans, most commonly to detect bleeding in and around the brain, blood clots in the lungs, and abnormalities in mammograms.
The U.S. Food and Drug Administration has now approved more than 500 medical devices (most of which are software packages) enabled by AI. The lion’s share are for radiology, where the technology helps doctors analyze MRIs, CT scans and X-rays faster and more accurately.
A clinical trial led by the University of Michigan offers a glimpse into how AI can complement human expertise. Surgical teams analyzed brain tumor tissue from 278 patients: Half of each sample was processed using AI, the other half by a neuropathologist.
Each method arrived at the correct tumor diagnosis 94% of the time. And each time one came to the wrong conclusion, the other got it right.
These sorts of results don’t suggest AI will replace human health care providers. But they do point to a future where we’ll see more “AI plus doctor” approaches to care. As a Harvard Medical School professor recently told The New York Times, “there are tasks that are better done with computers.”
Indeed, The New England Journal of Medicine published a special report on how AI chatbots could assist in the practice of medicine. The report found that the GPT-4 chatbot could produce accurate medical notes from the transcript of a patient visit; it could also extract information for prior insurance authorization and generate lab and prescription orders.
Another recent article told of a rheumatologist who had GPT-4 write an appeal to an insurer that had denied coverage for a patient’s drug. After receiving the letter, which would typically have taken the doctor a few hours, the insurer granted the request.
I think I speak for most physicians when I say that a future where AI can – consistently and accurately – handle tasks like these can’t come soon enough.
Do you have a health query for Dr. James? Email email@example.com and your question may be answered in a future column!