Ryan Radecki, MD MS @emlitofnote

FACEP FACEM AFAIDH. Emergency Medicine & Informatics. @AnnalsofEM Podcast & Journal Club. Columnist @ACEPNow. Speaker, writer, dad. @Stanford '00. 🇺🇸 in 🇳🇿 evidencetriage.com Christchurch City, New Zealand Joined June 2011

Tweets

23K
Followers

14K
Following

322
Likes

3

Ryan Radecki, MD MS @emlitofnote

5 days ago

A colossal distortion of flimsy biased data. A "stroke miracle" that ... isn't, but if you've invested £30m, you're not going to admit defeat .... #FOAMed evidencetriage.com/p/the-stroke-m…

1 2 5 687 2

Ryan Radecki, MD MS @emlitofnote

a week ago

Can't keep track of all the #ESCCongress articles? I've gathered most of the @NEJM publications under one umbrella for you. #FOAMed evidencetriage.com/p/its-european…

0 3 6 885 0

Ryan Radecki, MD MS @emlitofnote

a week ago

The hullabaloo about point-of-care troponin and decreased LOS misses the point – you can discharge any individual chest pain patient more quickly, but it just re-orders your work queue and the other patients wait a few minutes longer. #FOAMed evidencetriage.com/p/the-precario…

0 0 2 258 0

Ryan Radecki, MD MS @emlitofnote

2 weeks ago

The analysis of cost-savings is not quite as interesting as the care journeys of those who start their primary care journeys either through "traditional" access or "digital first" via an app. Either pathway seems fairly sticky! #FOAMed evidencetriage.com/p/where-do-dig…

0 1 1 224 0

Ryan Radecki, MD MS @emlitofnote

3 weeks ago

Impressed by OpenEvidence's perfect score on the USMLE? Well, turns out an LLM can guess its way to 60+% correct on MCQ based off an entirely fake organ in the human body just off subtle cues in answer construction. So, don't be as impressed! #FOAMed evidencetriage.com/p/are-llms-che…

0 0 2 320 1

Ryan Radecki, MD MS @emlitofnote

3 weeks ago

Epic is integrating a whole host of AI into their EHR – the most interesting of which is their attempt to build a generalized prediction engine off Cosmos using the transformer architecture. Not perfect, but an interesting start! #FOAMed evidencetriage.com/p/hype-vs-real…

1 2 3 483 1

Ryan Radecki, MD MS @emlitofnote

3 weeks ago

From today's AMIA SmartBrief: "Despite expectations that artificial intelligence would reduce work hours, it has primarily increased efficiency and productivity, driven by anxiety, competition and corporate greed, Boris Shpylchak writes." 🎯 psychologytoday.com/us/blog/the-ps…

1 0 1 242 0

Ryan Radecki, MD MS @emlitofnote

3 weeks ago

"Across all simulations, 56% of users failed at least once by clicking on an embedded phishing URL. Additionally, 25.9% of users failed at least two phishing simulations, 9.8% failed at least three, and 3.5% failed at least four exercises; one user failed every single time."

Ryan Radecki, MD MS @emlitofnote

3 weeks ago

0 2 4 2K 0

1 1 2 957 0

Ryan Radecki, MD MS @emlitofnote

3 weeks ago

There's not much apparent benefit to that obnoxious phishing training to which we're all subjected. And, wow, the secret to breaking into our hearts is apparently changing the dress code or the vacation policy, not Docusign or open enrollment. #FOAMed evidencetriage.com/p/does-phishin…

0 2 4 2K 0

Ryan Radecki, MD MS @emlitofnote

3 weeks ago

What *is* the normal range for "black blood cells"? Oh, thank you LLM, I was at a loss. #medsky #MLsky evidencetriage.com/p/medical-llms…

0 1 1 270 0

Ryan Radecki, MD MS @emlitofnote

4 weeks ago

Yes, LLMs have a "none of the above" problem in question bank challenges – but, well, so do humans. And don't call it "reasoning", but the frontier models do a better job of self-auditing and verification, and that's important. #FOAMed evidencetriage.com/p/the-llm-none…

0 1 1 295 1

Ryan Radecki, MD MS @emlitofnote

4 weeks ago

Neffy? Blujepa? Journavx? Atzumi? Are these weird AI hallucinations or real drugs you're going to be asked about in your emergency department? Latest in @ACEPNation ACEPNow: #FOAMed acepnow.com/article/whats-…

0 0 1 286 1

Ryan Radecki, MD MS @emlitofnote

a month ago

You can give alteplase beyond the 4.5 hour window when patients have favorable perfusion imaging – but that does imply some folks inside the window would have unfavorable perfusion, wouldn't it? #FOAMed evidencetriage.com/p/is-the-throm…

1 1 5 386 0

Ryan Radecki, MD MS @emlitofnote

a month ago

Publishing on clinical applications of AI? If you’re building a “chatbot”, there’s a new publishing reporting guideline to consider. #FOAMed evidencetriage.com/p/need-another…

0 0 0 228 0

Ryan Radecki, MD MS @emlitofnote

a month ago

Boarding in U.S. emergency departments – in anecdote and in real data – continues to rise. Prolonged ED stays are well-established contributors to adverse events. Another creeping "status quo" that simply isn't acceptable. healthaffairs.org/doi/abs/10.137…

1 4 4 525 1

Ryan Radecki, MD MS @emlitofnote

a month ago

I find this little experiment more amusing than most – why did they have to train their model to like owls? But, the underlying principle – that increasing seemingly-unrelated weights have wide-ranging effects – is quite fun. #notFOAmed #justfun evidencetriage.com/p/hypnotizing-…

0 1 1 1K 0

Ryan Radecki, MD MS @emlitofnote

a month ago

If you've ever wanted to move to New Zealand – or even just test it out for a start – you can't get much better an option than Christchurch. Come join our EM family and see for yourself. cdhb.careercentre.net.nz/Job/Emergency-…

2 1 5 430 1

Download Image

Ryan Radecki, MD MS @emlitofnote

a month ago

You could be all fancy with ECG-enabled smartwatches or monitoring platforms to try and catch atrial fibrillation in the community ... or you could just have an app that listens to your heartbeat jumping around. #FOAMed evidencetriage.com/p/just-regular…

0 1 2 277 0

Ryan Radecki, MD MS @emlitofnote

a month ago

Epic's pediatric sepsis model as failing external validation being rolled out as a "buyer beware". Bonus points for the authors correctly identifying the circular logic behind "improving sepsis mortality" resulting from over-identification. #FOAMed evidencetriage.com/p/the-predicti…