@KentMcCannMD Profile picture

Kent McCann, MD

@KentMcCannMD

ER/Palliative Physician | AI & Data Science | MedEd | Wellness Just a physician larping as a data scientist.

Similar User
Brian photo

@SoaringB_Rad

retired Pharmacist photo

@drugmancefn

Nicole Hatchard, DO photo

@nicolehatchard

Mihai Tudor photo

@RealMihaiTudor

Elizabeth Marras photo

@e__marras

Amber 💖 MA, LPC, BCC photo

@purposedsailor

Manuel photo

@Manuel6484

Gillian Mathews, D.O. photo

@GMHMathews

𝐀𝐡𝐦𝐞𝐝 𝐍𝐚𝐡𝐢𝐚𝐧, 𝐁.𝐒., 𝐌-𝐈𝐈 photo

@iamahmednahian

Courtney Ferguson photo

@cfergu99

Ryan James O'Sullivan photo

@rynetry

Maryam Khan photo

@maryamb_k

Johnny O'Connor photo

@lifesizemind

Austin Shokraeifard, PGY-2 photo

@AustinShok

Dexter Nwachukwu, DO, MS photo

@NwachukwuDexter

Is it standard practice for conferences to not publish a list of speakers/presenters along with the names of the presentations?


Kent McCann, MD Reposted

It's not necessarily an either/or question: LLMs are great for prototyping with no data, and you can then use them to create data for task-specific language models that are more accurate, much smaller and faster, and that you can run privately and efficiently in-house!

This is what I am wondering about, too. @spacy_io, any guide on when to use spacy and when to use LLMs, e.g. for classification?



Anyone familiar with #medspacy? Need some help using the DocConsumer on a previously processed docs object while tracking which doc each item in the resulting dataframe comes from.

Tweet Image 1

Kent McCann, MD Reposted

RAG vs. Fine-Tuning Cool report discussing the tradeoff between RAG and fine-tuning when using LLMs like Llama 2 and GPT-4. It performs a detailed analysis and highlights insights when applying the pipelines on an agricultural dataset. Here is a figure showing the pipeline…

Tweet Image 1
Tweet Image 2
Tweet Image 3

2024: the year anyone can write programs thanks to AI. Some of the coolest ideas are going to come from hobbyists with no coding background.

This morning I created a working SQL database from scratch and coded the JavaScript to fetch the data from a web form I created yesterday, to automate creating PDF documents with python. I’m a non technical person with no experience but it works🤯



We shouldn’t wait for corporations to dictate how #AI can alleviate burdensome administrative tasks in medicine. We should discover how ourselves.

Tweet Image 1

Listen to this podcast if you’re trying to learn how transformers (and LLMs like GPT) work It’s technical, but they break it down well. @superdatasci superdatascience.com/podcast/techni…


Google, using a fine tuned PaLM-2 model with self-generated training dialogue and Chain-of-Reasoning online inference, outperformed PCPs in diagnosis, and on quality of history taking & communication skills. #AIinMedicine arxiv.org/abs/2401.05654

Tweet Image 1
Tweet Image 2
Tweet Image 3

Great use of new custom GPTs. We’re going to see some major changes to medical education over the next few years.

Med Students!🚀 I made a ChatGPT tool to help with USMLE studying. It is called "Anki QuickCard". Just screenshot a question from UWrld, AMBOSS, etc, paste into the chat, & it automatically generates 2 Anki cards + key points instantly. Its saved me so much time. #MedTwitter

Tweet Image 1


"'creative thinking and empathy, will be in high demand,' says Saidov. 'Proficiency in using and understanding AI technology' should also be part of their portfolios." The same will be true for health care workers. forbes.com/sites/joemcken…


Can you diagnose TB from the sound of someone's cough? #AI can.

Screening of tuberculosis in Nairobi via a smartphone passive cough and deep learning #AI science.org/doi/10.1126/sc… @ScienceAdvances

Tweet Image 1


This @JAMAPediatrics article is making some news headlines recently: jamanetwork.com/journals/jamap… And the headlines are catchy. They're all some variation of "GPT missed 80% of diagnoses in pediatric cases" But there are a few things worth noting here. •This study used GPT 3.5,…

Tweet Image 1

Why is it that most medical CME courses ignore the past few decades of research on education and learning?


“AI is becoming an undercover operating system for professionals, particularly when it comes to using the technology for research and idea generation.” “Soon, workers will not even realize they are using AI to get their desired results.” forbes.com/sites/joemcken…


There's a huge benefit for LLMs in healthcare that I don't see anyone talking about: converting unstructured data (such as written notes) into structured formats. In a single prompt, ChatGPT converted a free-text RoS into this JSon representation. #MedTwitter #AI

Tweet Image 1

Almost 1/3 of academic physicians report "moderate" or higher intention to leave their current institution within the next 2 years. Burnout and lack of professional fulfillment were two associated factors. We need serious change in academic medicine.

1/ Welcome to Edition 181 of West’s Well-Being Wednesday! Today, a 🧵 on physician #wellbeing and intention to leave their current institution, from a recent paper @JAMANetworkOpen #wellbeingwednesday #medtwitter #meded jamanetwork.com/journals/jaman…



Loading...

Something went wrong.


Something went wrong.