Do you know this old saying?
A terrible thing happens when someone doesn’t live up to their potential: Nothing
Remarks by AMA Immediate Past President Jesse M. Ehrenfeld, MD, MPH, suggest that this idea could also apply to augmented intelligence (AI)—often called artificial intelligence—in health care.
“As I travel the country, there's so much uncertainty about how do we get this right,” Dr. Ehrenfeld said at the inaugural RAISE Health Symposium in Palo Alto, California. “As a practicing clinician—I was in the OR yesterday and saw 11 patients—I see that there's so much opportunity to get this wrong.”
Dr. Ehrenfeld was interviewed at the event by Fatima Rodriguez, MD, an associate professor of cardiovascular medicine at Stanford University School of Medicine. RAISE Health—which stands for “Responsible AI for Safe and Equitable Health”—is a joint initiative between Stanford Medicine and the Stanford Institute for Human-Centered Artificial Intelligence seeking to guide “the responsible use of AI across biomedical research, education and patient care.
“There's so much excitement, there's so much enthusiasm, and we're at the peak of the hype curve,” said Dr. Ehrenfeld, an anesthesiologist and health informaticist who served as AMA president through June. “But, if we don't have certain conditions on the regulatory side, on the liability issue, and a shared awareness around how we have ethical principles that can lift up equity as a foundational way to make things better—not worse—there is the potential for this to go horribly off the rails.”
Dr. Rodrguez asked him about the recent AMA survey on physicians’ use of AI in health care, and Dr. Ehrenfeld noted that the results showed that 41% of U.S. physicians “are equally excited as they are terrified about AI.”
The survey also found that 38% of physicians are using AI in their practices—but mostly for “unsexy back-end office stuff,” not for clinical applications.
“It's supply-chain management, it's scheduling, it's fighting for prior authorization approvals with all the third-party payers,” Dr. Ehrenfeld said. “Those systems are here, they're being embraced, and they're certainly useful to some degree.”
From AI implementation to EHR adoption and usability, the AMA is making technology work for physicians, ensuring that it is an asset to physicians—not a burden.
Learn from history
Developers of AI tools and other digital health applications must pay attention to—and avoid repeating—the mistakes made during the rollout of electronic health records, Dr. Ehrenfeld said. Until recently being overtaken by prior authorization requirements, EHRs were the No. 1 driver of dissatisfaction among physicians, he noted.
“A lot of technologies promised that they were going to make our lives better, make things more efficient, give us more capacity, but they didn't really pan out,” he said, adding that this still holds true today.
Dr. Ehrenfeld described how he recently saw a surgery resident having to help an 81-year-old patient sit up and then lean over her gurney to complete an electronic surgical consent form on a mobile “computer on wheels.” It would have been so much easier to just use a signature pad, he noted.
“It's just another very simple, obvious example of the technology, the deployment, didn't really match the workflow and just caused frustration,” he said. “Obviously, we can't let that happen over and over with these AI deployments.”
The clumsy rollouts of EHRs and other health technology usually have one thing in common.
“The fundamental mistake was we didn't have the physician voice present throughout the design-development-deployment cycle of EHRs,” Dr. Ehrenfeld said.
“I see this with entrepreneurial companies where there's a physician who might be involved, but it's an afterthought—they're not really driving the development of the solution,” he added. “That's a problem. “We can't let that continue to happen.”
For successful adoption, digital tools must answer four basic questions physicians have:
- Does it work?
- Will it work in my practice?
- Will I be paid enough to cover the cost of investment?
- If something goes wrong, will I be liable?
Two recent developments have created cause for concern regarding the last question.
The U.S. Department of Health Human Services Office of Civil Rights issued a rule concerning the nondiscrimination provision in Section 1557 of the Affordable Care Act (PDF) that could impose penalties on physicians if they rely on algorithm-enabled tools that result in discriminatory harms.
Likewise, the Federation of State Medical Boards issued a set of principles declaring physicians to be liable for harm caused by algorithm-enabled tools.
Fighting for Physicians
Get updates on how the AMA is fighting for physicians on critical issues—delivered to your inbox.
“The AMA believes that accountability should rest with those in the best position to know the potential risks of the AI system and to mitigate potential harm, such as developers or those mandating physician use of the AI tool,” Dr. Ehrenfeld said in a social media post. “Transparency plays a big role here and the AMA believes physicians should not be held liable when information about the quality or safety of the AI system is unknown or withheld.”
Previously, the AMA released its own advocacy principles (PDF) that address the development, deployment and use of health care AI.
Reframing health care work
Dr. Ehrenfeld covered similar ground while giving a keynote address at the recent Consumer Technology Association’s HeathFuture Summit in Chicago. There, he discussed health workforce shortages, patient access problems and how AI and other technology can help.
He said the health care industry needs to “reframe the work,” and to do so it must “lean on technology.” That having been said, technology should not replace doctors, nurses or pharmacists.
Concerns about the patient-physician relationship and patient privacy need to be addressed and physicians need clear information about what algorithms are doing and how they’re doing it.
“We must demand transparency,” Dr. Ehrenfeld said, adding that, if he is an operating room where an AI algorithm is controlling a patient’s ventilator, he needs to know “How do I hit the ‘off’ switch?”
For more discussion from Dr. Ehrenfeld on these issues, watch his appearance on a recent episode of “AMA Update.”
Learn with the AMA about the emerging landscape of augmented intelligence in health care (PDF).
