The future of AI in medicine is osteopathic

Since Andrew Taylor Still, DO, MD, flung to the breeze the banner of osteopathy in 1874, dreaming of a different and better way to practice medicine, our profession has been interested in the future and not afraid to walk on the cutting edge. We are the innovators in medicine; the first to recognize that form is integral to function and the necessity to treat the whole person—mind, body and spirit—when we heal illness, not simply focusing on the symptoms. We dared to practice in a way that diverged from the prevailing theories of the time. To this day, we are not afraid to embrace new theories and methodologies.

If there was one big takeaway from OMED 2023, it’s that artificial intelligence (AI) and other emerging health care technologies are on everyone’s minds. Over the past 12 months, new players such as generative large language model (LLM) AI have exploded onto the scene, capturing the public attention and igniting imaginations across a wide array of disciplines for their tantalizing ability to make earth-shattering changes to how we live and work.

Naturally, health care is no exception; in fact, AI has the potential to improve the practice of medicine more than perhaps any other profession by directly addressing staffing shortages, reducing documentation/administrative burden and physician burnout, as well as augmenting patient care by assisting in diagnosis and treatment decisions like a pocket curbside consultant.

An uphill battle

One of the unfortunate things that sets health care apart from other professions is that in our line of work, the technology we use often tends to be an obstacle to what we’re trying to achieve. An enduring problem with health care today is that the system is set up to disrespect our time. Everywhere you look, your time as a health care professional is disrespected at new levels. A recent study showed that for the average physician, every hour of patient care is coupled with two hours of administrative time. Another study showed that to complete all the required and recommended preventive services in the primary care setting in one workday, a single physician would require 26.7 hours.

Even in the best-case scenario, with a fully staffed health care team, it would still take many hours to do everything that most of us are being asked to do with far less support. At OMED23, a quick visual poll of the mainstage room yielded the staggering on-the-spot statistic that the majority of the physicians in a room of several hundred attendees admitted to spending several hours each workday doing charts at home after hours. There is a word coined for this all-too-common phenomenon: “Pajama Time.” This is not the way that medicine was meant to be practiced, and this is not our fault.

Glimmer of hope

This is where AI comes in. What was initially a promising pipe dream to answer some of health care’s most vexing problems has since evolved into a suite of real, practical technologies with multiple competing vendors and specialized applications. Ambient dictation—which might also be called a “virtual medical scribe”—involves using natural language processing (NLP) to transcribe a natural conversation between a patient and physician, then having an LLM program create a SOAP or progress note from the text, completing minutes-to-hours of manual charting in mere seconds.

Optical character recognition (OCR) is a technology that has been used by companies like TurboTax for years to digitize printed documents. OCR is another technology that can be plugged into an AI program, combining the recognition of printed text with the ability to analyze it. Imagine having an automated scanner review hundreds of pages of paper medical records and produce a summary of the most critical details, reconcile historical details with your current charts or find that one elusive echo or DEXA result you need.

This is not science fiction; there are dozens of companies that offer products which do this today. Newer technologies are pushing this one step further, adding LLM overlay to clinical decision resources and existing databases of guidelines and peer-reviewed studies to produce “pocket curbsides,” a physician-level artificial intelligence that can answer your half-baked UpToDate queries with the skill and patience of a seasoned consultant.

Large EMR companies like Cerner and Epic are already testing and employing AI that can analyze your note or clinic encounter, review the guidelines and suggest appropriate orders as well as schedule follow-up appointments and preventive screenings. In our practice, we are using AI ambient dictation, generative AI for text creation and a free, nonprofit LLM called Open Evidence (trained solely on peer-reviewed published papers) for research and guideline review.

Addressing concerns

Naturally, the breakneck pace of these innovations has evoked some concerns and questions.

It is vital to note that while the independence and automation displayed by AI is astounding, these technologies are not meant to replace physicians. Like autopilot in an airplane, the osteopathic physician is still in the pilot’s seat between the technology and the patient. Rather, AI algorithms are tools that we can use to do our jobs better, faster and with more joy.

Likewise, some physicians that are more cautious about AI have shared worries that AI will take away from already limited face time with patients. We do not see a mechanism for how this is likely to happen. On the contrary, AI should be able to reduce the amount of time physicians spend on non-patient care tasks such as documentation and should free the physician to spend more time with patients, not less.

We currently have a critically important window of opportunity to lead the development of this technology. AI only knows what and how it is trained to think about the world, and the intuitive conclusions that make it so powerful are still drawn from that foundational framework. Training an AI algorithm is not dissimilar from teaching a trainee or, in some ways, even parenting a child. While AI programs are by nature machine-like and robotic, we have the ability to inject humanity into them.

A team effort

How do we do this? Get involved! There are opportunities at every level.

If you are a student or trainee, seek out educational opportunities. Digital medical practice is already being taught in colleges of osteopathic medicine (COMs); if your COM is not teaching this material, ask your administrators to explore incorporating this into the curriculum. Make connections with other students. Form an interest group.

If you are an attending physician, seek out continuing medical education (CME) in this topic and try to read as much as possible. Major journals such as the Journal of the American Medical Association and the New England Journal of Medicine have created open access AI-specific sub-publications, with new AI research and commentaries rolling off the presses daily.

Additionally, as an attending, you have an opportunity to be an advisor and early practice adopter of emerging technologies. AI and health care information companies are rolling out new tools and many are currently looking for physicians to participate in beta trials. These trials usually allow you to use the software for free and provide feedback. Reach out to representatives from these companies and pay attention to calls for volunteers/testers.

Furthermore, federal agencies and regulatory bodies are beginning to craft policy governing the use of AI in health care, and this is the time osteopathic voices need to be heard in official comments. Keep abreast of policy efforts and position statements from any organizations/specialty societies you may be affiliated with and volunteer to lend your voice.

Osteopathic medicine is the most well-positioned profession in the world to guide how AI will be employed in health care. Our values of personalized, whole-person care can partner with technology to ensure the greatest benefit to both patients and physicians. However, if we stay complacent—or worse—fail to embrace AI due to fear of change, the outside forces which have been driving the progressive, systematic dehumanization of health care over the past decades may take the opportunity out of our hands and dictate how the technology will be employed for us.

Related reading:

The pros of artificial intelligence in health care

Artificial intelligence: Why it doesn’t belong in medicine