Immersing new docs in all things AI | Forecasting without hyping healthcare AI | And more AI doings of note

Show Kimberly Lomis, MD, a residency program light on AI instruction, and Dr. Lomis will show you a syllabus sorely lacking in the new basics of physician training. 

“If you’re worried that AI may amplify bias, or you’re worried that residents and fellows may be deskilled because of using these tools,” says Lomis, the vice president of medical education innovations for the American Medical Association, well, “there’s only one way” to head off such scenarios from materializing in real-world care settings. That way involves education, preferably during the newish physician’s formative years. Lomis and other healthcare AI proponents make their case for early trainee immersion in AI—its good, bad and potentially harmful aspects—in a featured news item posted Dec. 18 at AMA’s website. Here are highlights from their argument, posted Dec. 18. 

• Not long ago, the conversation around AI in postgraduate medical education seemed to favor downplaying AI altogether. Sometimes the instinct to ignore stemmed from a fear of inadvertently promoting an unproven technology, Lomis suggests. “We’ve moved past that, fortunately,” she tells AMA news writer Georgia Garvey. “[N]ow we’re in a phase where we recognize that it’s indeed those concerns that create the need to teach people.”
   
• It’s critical that program directors understand how AI might shape the clinical skills of residents and fellows. That’s the view of Bruce Levy, MD, director of the clinical informatics fellowship program for the Pennsylvania-based Geisinger College of Health Sciences. Should trainers and mentors be sure residents and fellows know how to write a clinical note, Levy asks, before they let an algorithm help with the task? “When is the appropriate time to introduce the tool in training?” he continues. “How much should they use it? These are all really important questions that we’re going to have to take time to better understand.”
   
• It wouldn’t be optimally edifying to warn of AI’s weaknesses without detailing its strengths. Lomis says she’s encouraged when residents tell her how they’re already leveraging AI for appropriate digital assistance. “Very few are using it to get a quick answer,” she reports. “Most are using it to offload administrative-type tasks, similar to how practicing physicians are doing it, or as a sounding board to test ideas and deepen their understanding.” She also finds many physician trainees smartly tapping AI to help ward off burnout. “If we can offload tasks that are not useful in terms of the learning part of their educational process, that can free time for other activities—including rest.” 
   
  • Read the rest and note the links to educational AMA resources here.

Peak AI hype seems to have passed. Sobered by reality, formerly breathless futurists can now get a fair hearing when they calmly state the technology really will transform medicine. 

They’ll just have to change “AI will replace doctors” to “AI will help clinicians humanize healthcare.” That’s one takeaway from a thought-provoking article posted online Dec. 18 at City Journal. “It will not be a simple task to integrate the strengths of AI with the skills and judgment of human medical workers,” writes the author, James Meigs, who was never an AI hypester and is, in fact, a former editor of the practical periodical Popular Mechanics. “But if we can fully exploit AI’s benefits, learn to temper its risks—and head off efforts to overregulate the AI revolution before it starts—a more effective, more humane vision of healthcare is within reach.” Here are more pearls from the piece. 

• For now, the benefits of AI tools in medicine appear to dramatically outweigh the risks. “Even if the grandiose predictions of future AI capabilities never materialize, today’s real-world AI tools already show the potential to reshape medicine for the better,” Meigs observes. “In many cases, those benefits will prove lifesaving.”
   
• The 2022 release of OpenAI’s ChatGPT platform—soon followed by Google’s Gemini, Meta AI and other LLMs—opened up a new vista of possibility. “I talked with one oncologist who routinely uses ChatGPT to draft letters to insurance companies seeking preauthorization to cover the cost of expensive new drugs,” Meigs shares. “The process saves him hours each week. That’s time he can use talking to patients instead of insurance companies.”
   
• Virtually everyone who has studied LLMs in medical applications agrees that this technology will be a force multiplier for overworked doctors and nurses. “AI platforms should also help level the playing field for rural or underfinanced hospitals that lack cutting-edge medical expertise,” Meigs notes. “First, though, the bugs need to be discovered and worked out.”  While the finetuning is happening, however, lawmakers have to resist the temptation to address various AI limitations through premature regulation, Meigs writes. “The best way to avoid AI pitfalls is to continue the kind of research that Eyal Klang, MD, [of the Icahn School of Medicine at Mount Sinai] and others are pursuing,” he asserts. “Then healthcare organizations must develop and document—and continuously monitor—best practices for using AI.”
   
• Remember: Machine learning systems can see micropatterns in medical images and records that no human expert could realistically be expected to discern. Thanks to this kind of super-augmentation, “AI will help doctors detect—and someday, perhaps, prevent—diseases like Alzheimer’s and cancer years before they become clinically observable.”
   
  • Read the whole thing.

Also worthwhile:

• Report: Health systems see promise in generative AI for documentation and revenue cycle accuracy, but adoption gaps persist (Healthcare Financial Management Association and Akasa) 
   
• AI’s job impact: Gains outpace losses (IT & Innovation Foundation) 
   
• Artificial intelligence detects early signs of aging from chest X-rays (Harvard Medical School) 

From HealthExec’s sibling news outlets:

• Radiology dominates FDA-cleared AI, but reimbursement lags far behind (Radiology Business) 
   
• AI flags hard-to-detect heart disease in seconds (Cardiovascular Business)