Tim Miller
2.3K posts
Tim Miller
@TMills
Researcher in clinical #nlproc at Children's Hospital Computational Health Informatics Program (CHIP) and Associate Professor at Harvard Medical School.
Boston, MA Katılım Mart 2007
191 Takip Edilen570 Takipçiler
Tim Miller retweetledi
Reminder to join us September 17th, at 12 EST for the next #RegSci forum! Dr. Tim Miller @TMills will be presenting on "Opportunities for Large Language Models in Regulatory Science"! #LLM #NLP #clinicalNLP
Register here: rb.gy/dxi1ma
Harvard-MIT Center for Regulatory Science@HarvardMIT_CRS
Next #RegSci Forum, on September 17th, will feature guest speaker Tim Miller @TMills on “Opportunities for Large Language Models in Regulatory Science”. Register here: rb.gy/psky6m
English
@nsaphra How do you judge a review to be chatgpt generated? It's easy to see at a population level but are there tells for individual reviews?
English
Relatedly, how do we reach out to our AC about a slop review (for neurips)?
English
You know, after all the panic around emergency reviews missing and the ACL reviewing pool diluting: I received more ChatGPT generated slop in my 3 NeurIPS reviews than in my 13 EMNLP reviews.
English
@Serena_pancakes Very excited for you, and looking forward to continued collaboration!
English
🌿Officially: I’m excited to join the DBMI at CU Anschutz this September as assistant professor! 🏔️
Can’t wait to see the sparkles fly when NLP meets cutting-edge healthcare and medical research at CU Anschutz. Look forward to new collaborations and innovations!
CU Department of Biomedical Informatics (DBMI)@CUBiomedInfo
Yanjun Gao, PhD, (@Serena_pancakes) will join our department as an assistant professor this fall! She will focus on developing foundational natural language processing technologies and conducting research on innovative AI tools in clinical settings. news.cuanschutz.edu/dbmi/yanjun-ga…
English
True test of loyalty for Indian Americans. The match up we dread the most 🙂
English
@barnfell I got the Slim fold wallet (I think micro soft shell) and it has held up well. My wife and friend bought one as well.
English
does anyone know how to buy a nice wallet that won’t fall apart within a couple years of daily use
English
@deliprao Also, as a mid or late career academic conferences are the most fun part of the job.
English
“Academia is so easy and relaxed” or is it?
Interesting cognitive bias here. Successful academics are usually high achievers who can handle a ton of stress. It’s easy to see a successful academic from the outside and assume they must have it easy because they carry their stress well.
yi@agihippo
Last week I dropped by a conference and I think academics and professors have a really good life not having to worry about anything. Academia is so easy and relaxed.
English
Tim Miller retweetledi
1/9 Excited to introduce the LCD Benchmark! A new benchmark dataset for 30-day out-of-hospital mortality prediction (predicting the future!) using long clinical documents. Designed for testing the capabilities of LMs on lengthy documents! medrxiv.org/content/10.110… #ClinicalNLP
English
Tune in for great sentences like, "The Comstock Act might be one of the biggest sleeper issues of this campaign."
Doni Holloway@doniholloway
Don't miss it: we're out with the latest episode of our #WITHpod2024 The Stakes series. This week, @JessicaValenti, founder of abortioneveryday.com, joins @chrislhayes to talk about the stakes for reproductive rights. “We’re already in a health care crisis, but it could get so much worse. And I don't know that voters completely understand just how much worse it could get," she points out. Hear the full convo wherever you get your pods: link.chtbl.com/withpod_tssg #WITHpod
English
Tim Miller retweetledi
#LLMs might help burnout by drafting responding to patient portal messages in #EHRs. But is this a low-risk efficiency aid, or do #LLMs impact clinical judgement in more meaningful ways?
We investigated this in our new study out in @LancetDigitalH: thelancet.com/journals/landi… 1/
English
Tim Miller retweetledi
NEW Comment from @dbittermanmd and colleagues: The effect of using a large language model to respond to patient messages. #LLMs
Read it here: buff.ly/3JzLT0t
English
I'm very sad to hear this. I read Dennett as an undergraduate at @MarquetteU and as a first year PhD student and it helped push me to study AI and language. I still talk about and think about his ideas all the time.
Paul Bloom@paulbloomatyale
Such sad news. He was a brilliant philosopher who had a profound influence on cognitive science. And he was hugely supportive to young scholars—including me, long ago. A great loss. dailynous.com/2024/04/19/dan…
English
Introducing the first post pandemic doctoral baby of our lab, Aviv Brokman, who just defended his dissertation. He did an amazing job. PhD #7 and my first statistics advisee. He will soon be on the job market. More later. Celebrations for now 🎉🥳🎊
English
Tim Miller retweetledi
Proud of @shan23chen presenting our work on synthetic clinical text generation to 🚀 accelerate clinical #NLP at #AMIA #IS24!
English
English
#PhDone
Excited to share that I defended my thesis this afternoon!
It was a wild, stressful, exciting, and unpredictable journey ‼️
Thank you @noemieelhadad for being there at every step!
English
Tim Miller retweetledi
It’s official on JAMIA! 🎈
After chatgpt being 1 years old and working with in context learning more. I think my points still stands firmly where LLM+prompting is fabulous, however the time investment and current performance gap is far from zero for domain specific tasks 😢
Danielle Bitterman, MD@dbittermanmd
#ChatGPT has taken #NLProc by storm 🤖. We investigated time, data, and performance trade-offs of ChatGPT family of models in 2 health NLP tasks beyond QA: health advice classification and causal reasoning. ⚕️ [1/6] arxiv.org/abs/2304.02496 @ShaaanChen @BrighamRadOnc @Bos_CHIP
English
Tim Miller retweetledi
💡🎉Learn more about the role of medical KGs in AI-augmented diagnostic systems via a graph prompting framework for LLM in our study.
👩⚕️👨⚕️ Featuring the first comprehensive human evaluation of LLMs for diagnostic safety. 1/🧵
#HealthAI #NLP #RAG
medrxiv.org/content/10.110…
English
Tim Miller retweetledi
GPT-4 gives medical advice that saves doctors' time but can be harmful #Echobox=1700033601" target="_blank" rel="nofollow noopener">newscientist.com/article/240204…
English
Tim Miller retweetledi
LLMs could help manage the EHR and reduce burnout, but need careful safety checks - want some evidence? #NLP #LLM #AI
✅Introducing OncQA, an open dataset with clinician responses & evaluations
@dbittermanmd @BrighamRadOnc @Bos_CHIP
📑 huggingface.co/papers/2310.17…
⚕️[1/6]
English
👋I boost a lot of job opportunities on here and now it's time to boost my own! I'll be arriving at #Stanford in fall of 2024 and I'm looking for awesome people to help me figure out language. 🧵👇
English