There is a lot of buzz about our new paper in Nature Medicine on the effects of LLMs (GPT-4) on physician management reasoning! I had TONS of fun working on this -- but what it MEANS requires some unpacking.

A 🧵⬇️

https://x.com/EricTopol/status/1887153992976765023

First, the main result: GPT-4 DID improve physician medical management in complex cases with no right answers (though not above the performance of the LLM by itself).

Preprint out today that tests o1-preview's medical reasoning experiments against a baseline of 100s of clinicians.

In this case the title says it all:

Superhuman performance of a large language model on the reasoning tasks of a physician

Link:

A 🧵⬇️arxiv.org/abs/2412.10849 This is the FIRST TIME I have promoted one of our preprints (rather than the full peer-reviewed study) so caveat emptor. But I truly think our results have implications for medical practice so I wanted to get them out as quickly as possible.

Can GPT-4 solve really hard medical cases and come up with a good list of differential diagnoses?

@zahirkanjee @byrondcrowe and my study is out in @JAMA_current , and the short answer is, “Yes.”

But what does this all mean? 🧵⬇️

https://twitter.com/JAMA_current/status/1669366653300211712

First, the topline results – in running all of the post-2021 published @NEJM clinicopathologic conferences, ChatGPT got the final diagnosis in 39% of cases, and had the final diagnosis in its differential in 64% of cases.

Since my thread on the historicity of the exam can gained some traction, here's a reading list if you're interested in gaining perspective on the nature of clinical reasoning -- rather than "just so" stories about imagined halcyon pasts (the era of "the Giants") First a beautiful essay by Faith Fitzgerald (it's only two pages): bumc.bu.edu/facdev-medicin…

"What does curiosity have to do with the humanistic practice of medicine? Couldn’t it just convert
patients into objects of analysis? I believe that it is
curiosity that converts strangers

So many arguments about what's wrong with medicine today are predicated on imagined (and inaccurate) histories. Let's take some examples from my colleagues who imagine a "golden" age of the exam: The physical exam is barely older than modern diagnostic tests. For example, the neurological exam was developed in the 1890s, the same time as the x-ray. We were ALREADY performing blood spears and gram stains and checking hematocrits when the neurological exam was developed.

I want to keep highlighting some of the amazing speakers we have at the @iMedEducation #DigitalEducation2022 conference, held virtually on October 7th, and in person in Boston on October 8th!

Next up -- @AshleyGWinter!

cmecatalog.hms.harvard.edu/digital-educat… Everyone has a professional or educational message that we want to get out to the world.

@AshleyGWinter is an expert at education and advocacy for sexual health and sexual medicine. She is going to be sharing her insights about this journey for all of us!

Why are medical podcasts like @thecurbsiders, @BehindTheKnife, @emcrit, and @AFPpodcast so popular for learning? And who is making them? And can they be trusted?

We listened to (and coded) the top 100 podcasts on the Apple podcasts US medicine chart to find out!

A 🧵⬇️ There were 2⃣ inspirations for this study.

@ShreyaTrivediMD and I at @iMedEducation think that what makes digital education unique from eg an uploaded lecture on YouTube is that that it is produced as part of a virtual community of practice and not traditional institutions.

Almost exactly a year ago, I had a modestly controversial tweet about routine daily physical exams -- and about how we should probably spend more time actually talking to our patients daily rather than pretending to examine then.

https://twitter.com/adamrodmanmd/status/1295417747984261124

Well, now that angry tweet is a point-counterpoint-rebuttal series in @JHospMedicine!

The first piece is by me and @ShaneWarnockMD, and I cut right to the point: Routine daily physical exams in hospitalized patients are a waste of time.

🧵⬇️

or
journalofhospitalmedicine.com/jhospmed/artic…

My (preaching to the choir) 🔥 take: digital educational skills -- whether teaching on #MedTwitter, podcasting, or making videos -- are essential #meded skills for the 21st century. And we can teach these to future educators.

A Tweetorial🧵: Last year, @ShreyaTrivediMD @StaciSaundersMD and I at @iMedEducation started a curriculum to teach digital educational skills to our @BIDMC_IM residents.

We just published this article going over our curriculum and providing tips for you to do it too: pubmed.ncbi.nlm.nih.gov/34013623/

A proposal!

Are you:

✅A listener of @BedsideRounds?
✅A medical student at a US or Canadian medical school?
✅Interested in medical history, philosophy, and epistemology?

And then question 2:

✅Do you want to collaborate on a research project?

If so, 🧵⬇️ First, some details! Over the past six years, I've made a lot of podcasts (some of them better than others), and I know they're being used for teaching at medical schools across the country.

Why do we use godawful blue-background-with-bright-yellow-text for medical school lectures?

A 🧵on magic lanterns, darkrooms, path dependence, and “things we do for no reason”

👇 Projecting images is quite old, dating back to the 18th century with images painted directly on glass plates -- a magic lantern. During the Spanish Flu, the Surgeon General toured with a magic lantern with information on the pandemic.

@COREIMpodcast Interesting thread, though some very common misconceptions about "primum non nocere" are present here. First do no harm is not in the Hippocratic Oath at all, and the "non-malfeasance" present in oath would likely make most modern doctors squeamish. @COREIMpodcast It has diktats against performing abortion, and against physician-assisted suicide. Even the commonly cited "do not cut for stone" isn't because of preventing harm, but suggesting that a lithotomist do it.

(good translation here: nlm.nih.gov/hmd/greek/gree…)

It’s time for another #histmed Tweetorial -- this time I'm going to talk about the pesky definition of a fever, and where the 98.6 F average body temp came from!

Full disclosure: will use C AND F for temp, but no K or R. FYI this is a complementary Tweetorial to @tony_breu's amazing one on why we have night sweats

https://twitter.com/tony_breu/status/1127262873577766912

It's time for me to channel my inner @tony_breu -- which means it's Tweetorial time!

So let's talk about azotemia (elevated blood urea nitrogen) after an upper gastrointestinal bleed! It’s a well-known phenomenon on the medical wards that after an upper gastrointestinal bleed, the blood urea nitrogen will rise considerably more than the creatinine. In fact, it’s a common teaching “pearl”

Time for a Tweetorial! Though this will only be partially #histmed and mostly about philosophy. Inspired by @chrischiu -- so let’s talk about Occam’s Razor and Hickam’s Dictum! But before we get going, let’s start with a little pre-test. Case #1. A young man presents with acute onset of severe fevers and chills, rhinorrhea, headache, confusion, and neck stiffness. What does he have?

Hey #medtwitter, it’s time for another #histmed #FOAMed Tweetorial! I’m giving a couple of lectures this fall, and in the spirit of #FOAM I’m going to (try my best) to do a Tweetorial for each, so anyone can benefit/watch me flounder/vehemently disagree with me. So thank you to @BostonChiefs, and let's talk about semiotics and the development of the physical exam!