William F Parker, MD, PhD

That means so much coming from you @iwashyna.bsky.social !

We have a long way to go, but are proud of what we have built so far

Check out our website and GitHub!

clif-consortium.github.io/website/

A common longitudinal intensive care unit data format (CLIF) for critical illness research - Intensive Care Medicine Rationale Critical illness threatens millions of lives annually. Electronic health record (EHR) data are a source of granular information that could generate crucial insights into the nature and optim...

The common longitudinal intensive care unit data format (CLIF) for critical illness research is among the most interesting new collaborative projects in critical care

This is a team to watch #ICU #RealWorldData

link.springer.com/article/10.1...

Science is about testing hypotheses- many NIH funded grants will turn out to be a waste of money. But the ones that don’t lead to life-saving breakthroughs.

NIH study sections play an essential role ensuring the ROI on taxpayer $$ is as large as possible

Getting extramural grant funding from the NIH is incredibly difficult- you need be in the top ~10% of all scientists in the country

I agree we should study and improve the review process but as a taxpayer I’m glad there is intense competition for these funds.

Patients need AIs that provide life-saving clinical decision support to their doctors.

Patients don't need AIs that scour clinical documentation looking for ways to charge them more

New @annalsats.bsky.social

Lest a Smoky Haze of Doubt Suffocate Progress Towards Better Pulse Oximeters

This was fun to write with Vincent Ni @jlynch13.bsky.social @hauschildt.bsky.social @drtomori.bsky.social

doi.org/10.1513/Anna... PubMed: 39700509

#FixPulseOx #FixTheDamnMachines

Make you* more way more productive 🤦‍♂️

If you care about advancing healthcare access and outcome for marginalized populations, embrace AI now.

These tools are publicly available and customizable to make way more productive

My new year resolution is that every data and normative analysis in my quantitative bioethics lab will be assisted by a fine-tuned AI

Totally agree with @wfparker.bsky.social! #AI will be integrated in everything we do in medicine, so should get used to it now.

Using it as a #Co-Intelligence, as @emollick.bsky.social suggests, is most appropriate. These examples by Dr. Parker are great!

I'm crabby today but I wish we could have a little less pearl-clutching over "the children are citing AI hallucinations" and a little more over "tenured academics are citing documents that cite documents that cite documents that cite documents that don't support their claims."

Sister Generose Gervais, a longtime administrator of the Mayo Clinic, is often quoted as saying, “No money, no mission.” A hospital can’t care for patients if it can’t pay for itself. But, O’Donnell told me, “people forget the second half of what she said: no mission, no need for money.”

The Gilded Age of Medicine Is Here Health insurers and hospitals increasingly treat patients less as humans in need of care than consumers who generate profit.

www.newyorker.com/culture/2024...

Agree, but my claim is that the decision-making should be driven by the risks and benefits of antibiotic therapy for that particular patient (not external considerations)

Sure, as long as the patient’s best interest and outcome is categorically more important and reducing antibiotic days is a means to that end.

This means treating a patient in shock with only a 10% of infection with broad spectrum antibiotics, for example.

The value judgments in the #ADAPT trial design are more interesting than the results

I give broad spectrum antibiotics to almost every patient on a vasopressor in the MICU. It’s because I value human life orders of magnitude more than antibiotic “stewardship” in my decision making

This is just ludicrous misvaluation of the relative importance of one day of antibiotics vs a human life

Clear win for standard of care! #ADAPT

jamanetwork.com/journals/jam...

Can someone please explain to me: “where the noninferiority margin was set at 5.4%” for 28 day mortality?

If you can answer all of these questions instantly and full detail and explain them to a trainee, I applaud your excellence.

But honestly, I think in most cases ChatGPT would do a better job.

As educators and doctors, we are ethically obligated to embrace these tools 6/6

What is the evidence for IM epinephrine in asthma? 5/

can organizing pneumonia develop in patients with pancytopenia? 4/

how do the results of the CLOVERS trial inform central line placement practices for shock? 3/

Why do continuous nebs cause a lactic acidosis? 2/

Every critical care physician should be doing an AI powered search for every patient, especially if they are teaching trainees

I used to be ashamed when I did this, but I realized it was because of the myth of the omniscient doctor.

Here’s a thread of just a few I did last week in the ICU:

🧵 1/

As long as you engage intellectually with the output, it’s the same as spending a couple of hours researching the question. AI has already shown it can exceed the diagnostic ability of us doctors- so it’s something I feel ethically obligated to engage with and use

I actually think it has a potential to make the interaction a lot richer- I’ve thought a lot more about Jack’s situation and what actually makes something “cross generational” that I would’ve otherwise

lol do you ever google anything? Same LLM runs automatically. Is your plan to stop using the internet completely for vaguely specified environmental concerns?

AI will almost certainly have net positive environmental effects in the long run as it makes electric grids more efficient

ChatGPT responses

Okay couldn’t help myself…not sure this is AIs best

Oh yes it’s far from “perfect”…I think best way to think about it is as speeding up a task you otherwise could perform yourself

Just start using it… one thing that might help is start a conversation with ChatGPT about a topic you know super well