Home » Dr. Jayne » Currently Reading:

EPtalk by Dr. Jayne 10/10/24

October 10, 2024 Dr. Jayne 1 Comment

clip_image002 

Just in time for the winter respiratory virus season, the Centers for Disease Control and Prevention has released new respiratory illness resources. New Community Snapshot content shows viral activity in different ways, including overall viral activity, wastewater viral levels, and emergency department visits. The site is scheduled to be updated on Fridays, with additional data elements added over time, including hospitalization trends.

Although COVID is on the decline in my area, we are just starting to see an uptick in RSV and influenza, so don’t forget to wash your hands, stay home if you’re sick, and make sure that you’re up to date on appropriate vaccinations. Given the respiratory crud that plagues many of us during conference season, I made sure I’m current for both COVID and influenza before hitting the ground at HLTH.

From Foodie: “Re: dangers of AI in clinical documentation. Have you seen this piece about the dangers of AI-created recipes?” I hadn’t seen it, but appreciate the share. I enjoy cooking and difficult enough sometimes to get a recipe to turn out correctly even when it comes from a reputable and well-tested source, so I’m not usually a fan of recipes from food influencers. (I admit, however, that I recently transcribed a recipe for crumpets from an online chef, which resulted in the need to acquire crumpet rings, and at some point, I will be testing it out so I’m not going to say never on that one.) The article lists examples such as one where a Twitter user entered prompts that led a recipe generator to suggest mixing bleach and ammonia, which creates fumes that are incompatible with life. Food bloggers are understandably worried about AI competition and note that AI can’t explore food from a sensory perspective to determine whether the recipes it creates are good. Other recipe creators have the same concerns that many have voiced about AI, including lack of attribution when content is used to train a model and intellectual property concerns.

Speaking of food-related adventures, a Harvard medical student decided to become his own science project and consumed 700 eggs during a month-long experiment. Despite taking in a tremendous amount of cholesterol, his own cholesterol values declined during the month. It should be noted that the subject’s cholesterol values were good prior to the experiment and that he’s a young, otherwise healthy individual, which is not the case for anyone. You and I don’t hare a physician / patient relationship, but this doctor is telling you not to eat 24 eggs a day. The student embarked on the project to make a point about messaging around diets as well as to encourage greater research in the field of metabolic health.

Several recruiters have reached out to me in recent weeks to try to lure me back to the in-person adventures of the emergency department. I know from speaking with former colleagues that quite a few emergency physicians have hung up their stethoscopes in the years since the beginning of the COVID pandemic. Maybe it was the feeling that your hospital felt that you were expendable and the lack of personal protective equipment. Maybe it was the idea that you weren’t ever allowed to be sick yourself or that taking a day off was unfair to the team. Maybe it was also being expected to deliver primary care when you’re not trained to do so, and not having the resources that you need to feel like you’re doing the right things for the patients in your care. Working in an emergency department can be exhilarating, but it’s also incredibly stressful and physically and mentally exhausting. There’s always the risk that you’ll miss something.

Emergency medicine has long been a proving ground for data-driven approaches to care, and a recent article from the American College of Emergency Physicians looks at the role of triage in the care of emergency patients and if it can be improved with better use of data. The authors note recent studies that estimate triage errors to be as high as one in every three patients, with vulnerable populations being at the highest risk. They propose the creation of new data-driven approaches to patient complexity that can take into account the numerous data points that are being captured on patients when they present for care as well as their histories and other elements that might be available to clinicians. They propose expanding the use of AI to synthesize available data and provide individualized risk profiles for patients at the point of care, noting that such models have been in place for several years at some institutions. It will be interesting to see how these solutions are incorporated at smaller emergency facilities and especially at those in remote areas that don’t always have in-house physicians. If you have experience with these solutions, drop me a note.

I recently ran across this study that looked at adverse diagnostic events impacting hospitalized patients. Although it has the limitation of being done in a single location, it reveals some significant findings. Researchers looked at harmful diagnostic errors, which included delays, process failures, and issues with subspecialty consultation. They estimated that a harmful error happened for one of every 14 patients, with the majority of errors being preventable. Although the authors call for additional approaches for diagnostic error surveillance, I think this work should be a call to action for error prevention as well.

In research like this, general terms such as “errors” or “harms” mask what really happens to patients in these situations. The article makes it a bit more clear: minor harms had mild symptoms or short-term impacts, while the other end of the spectrum included major harms that could have led to lifesaving surgical or medical interventions, shortened life expectancy, permanent loss of function, or even a fatality. Diagnostic errors include failure to make a clear diagnosis, misinterpretation of laboratory or other tests, incomplete workups, and other scenarios where patients don’t get the care they need.

The authors note that incorporating artificial intelligence could be helpful for the detection of “complex patterns of risk factors and clinical events that represent markers of risk or suboptimal diagnostic processes.” Tools to help with these scenarios have been around for many years, but have been slowly incorporated by care delivery organizations due to cost, lack of perceived benefit, and willingness to tolerate a higher level of risk than may institutions hold today. I look forward to seeing more solutions implemented over the coming years and for researchers to be able to quantify the number of lives saved or functionality preserved.

Is your organization using AI or other solutions to reduce diagnostic errors? Leave a comment or email me.

Email Dr. Jayne.



HIStalk Featured Sponsors

     

Currently there is "1 comment" on this Article:

  1. Medical student who ate the eggs–much safer than this notorious student from early in the last century and a great saga from the history of medicine.

    I recall being told in a lecture to us surgery students while on block at Duke med school by Dr David Sabiston that young Werner proved placement in the heart by walking down the stairs, self inserted catheter in place, to the flouroscopy equipment.

    https://en.wikipedia.org/wiki/Werner_Forssmann

Text Ads


RECENT COMMENTS

  1. Per Larry's visionary vapor session, they're calling it New Millennium and completed it with 500 people prompting Oracle's proprietary AI…

  2. "If at first, you don't succeed? Try, try again! Then give up. There's no point in being a damn fool…

  3. Hah! Avanel, the dream of a CTO whose only claim to fame was creating a spaghetti of a patient portal…

  4. The starving masses, and thus the pizza man who serves them, have left the building(s).

  5. We are getting less than 40 hours of work from a large number of our K.C.-based EMPLOYEES. The parking lot…

Founding Sponsors


 

Platinum Sponsors


 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Gold Sponsors


 

 

 

 

 

 

 

 

 

RSS Webinars

  • An error has occurred, which probably means the feed is down. Try again later.