A recent Medpage Today article reviews a research article published in JAMA Internal Medicine. Nidhi Subbaraman, in a topic-related Wall Street Journal article, also reviews this JAMA article and several pilot AI doctor response programs currently undergoing clinical testing.
Both of these reviews indicate that AI-created responses to patient-driven online questions posed to doctors are viewed more favorably (by the patients) and rated as having greater empathy than the doctor’s responses.
I reviewed this evolving phenomenon in my previous article. I pointed out that AI has already been able to pass medical school admission exams, medical board certification exams, and state medical licensing exams. AI was also found better at interpreting complicated imaging studies and tests and noted as better at finding previously missed diagnoses than acknowledged medical experts within the fields thus far investigated.
Both of these newer articles also stress that AI might be able to intervene in the steadily worsening burnout and overwhelming task burden increasingly facing physicians and other health care workers.
It is now increasingly noted that physicians, overburdened and exhausted with work-related tasks and disruptive schedules, are burned out and depressed leading to, among other issues, decreased empathy for their patients as well as increased medical errors. However, contrasting studies have shown that “increased physician empathy is associated with various improved health outcomes.” Again, these recent AI delivery studies and pilot projects emphasize that AI is currently perceived as better at delivering empathetic responses to patients.
I also pointed out in my previous publication that as the shortage of doctors and nurses continues to accelerate (between 4 to 5 million health care workers have left health care in the last two years – around 20 percent of the current U.S. health care labor force), shows no signs of slowing, and continues to go unanswered by political and health care leaders. The shortfall in physicians alone in the U.S. is projected to be between 54,000 and 139,000 by 2033, with over 50 percent of currently practicing physicians already over 50 years old.
As many health care executives and corporate leaders increasingly face worker push-back, experience rising unionization efforts, and see accelerating departures out of their now more frequently burnout-promoting, hostile, and dangerous health care workplaces (up to 75 percent of all U.S workplace assaults now occur in health care workplaces), these leaders counter by pointing out the worsening shape of hospital finances and that many hospitals in this country are now closing in great part due to inability to maintain adequate numbers of physicians, mid-levels, and nurses to keep essential revenue-generating service lines open.
Thus, despite multiple cautionary warnings of rushing into AI-driven solutions, in this steadily degenerating health care workplace climate, and in an era where leaders of increasingly-corporate mega-health care systems have been noted to state that “one of the biggest obstructions to profitable health care is doctors and nurses,” pathways for utilizing AI to address expanding needs — including this increasing shortfall in physicians, nurses and other health care workers in various capacities — are being aggressively explored.
If successful, AI-driven solutions could allow health care workplace leaders to address the exit of human health care workers while, with each step, reducing efforts and the perceived significant expenses to improve safety conditions in the workplaces themselves, and thus avoid having to address the underlying conditions leading to increased human health care worker departures.
Would patients actually find AI doctors truly to their liking? These recent studies suggest that they might, especially in early adoptive areas such as telemedicine.
Though perceived by some as profoundly expensive, wouldn’t it be more expedient to fix the issues currently plaguing our health care workplaces and workers? There are already pathways described by which to do so.
Could we eventually see a brave new world where AI and physical tasks drive medical decision-making and care efforts are largely performed via robotic efforts? Thus it would largely eliminate the extensive costs of managing (herding cats) and underwriting financial, human health care worker efforts.
Harry Severance is an emergency physician.