Posts by Julian Wolfson
Graphic featuring a photo of award winner Leif Verace and the title of his paper, "A Marginal Approach to Longitudinal Function-on-Function Regression"
So proud that UMN Biostat & Health Data Science's own Leif Verace received the John van Ryzin Award for best student paper at ENAR. Congrats Leif (and advisor/co-author Dr. Erjia Cui)!
powerful detailed account of ICE captivity
"To put things into perspective: I had a Canadian passport, lawyers, resources, media attention, friends, family and even politicians advocating for me...Imagine what this system is like for every other person in there"
www.theguardian.com/us-news/2025...
“How bad can it be to have an anti-vaccine zealot leading our health agencies?” 🤦🏻♂️🤦🏻♂️🤦🏻♂️
MINNESOTA
NIH Awards Funding: $718 M
Jobs Supported : 7,992
Economic activity supported: $ 1.74 B
www.unitedformedicalresearch.org/nih-in-your-...
Welcome to The Good Place! (ok, maybe not an ideal reference)
🚀 {unhcrthemes} 0.6.3 is out!
The latest version of our R package for creating UNHCR-styled charts is now on CRAN. 🎨📊
🔗 Check it out: cran.r-project.org/web/packages...
(1/3)
I kind of like it as is; a nice clock that just so happens to be a clever reference
There are post-doc openings for the @minnpop.bsky.social Training Program in Population Health Science. Apply by the end of January if you're interested and eligible! And spread the word please!
pop.umn.edu/training/pos...
Receiving a "Model Release" (the Zoolander kind, not the glm kind) form from NCI definitely wasn't on my bingo card for today.
Another edition of Shiny Source Code Explained ✨🤓 And this time it is all about the code behind the new busy indicators that were added in Shiny 1.9 (July 2024)!
hypebright.nl/en/shiny-en/...
Enjoy!
#rshiny #rstats
Quarto 1.6 is out! 🎉 Download it here: quarto.org/docs/download/
Quarto 1.6 supports unified branding across formats, updates to RevealJS, a new shortcode to reorder content, a landscape page block, and more. Blog post: quarto.org/docs/blog/po...
Have you asked your local barista?
Diagram of multiple imputation produced by ChatGPT, consisting of meaningless shapes, arrows, and incomprehensible text.
ChatGPT "explains" multiple imputation 🧐
Feel-good story? Check.
Public health connection? Check.
Cute dog? Check.
Awesome person doing inspiring things? Check and check.
www.cbsnews.com/minnesota/ne...
Key features of ‘tidyprompt’ are: tidy prompting: Quickly and elegantly construct prompts for LLMs, using piping syntax (inspired by the ‘tidyverse’). Wrap a base prompt in prompt wraps to influence how the LLM handles the prompt. A library of pre-built prompt wraps is included, but you can also write your own. structured output: Extract structured output from the LLM’s response, and validate it. Automatic retries with feedback to the LLM, if the output is not as expected. reasoning modes: Make your LLM answer in a specific mode, such as chain-of-thought or ReAct (Reasoning and Acting) modes. function calling: Give your LLM the ability to autonomously call R functions (‘tools’). With this, the LLM can retrieve information or take other actions. ‘tidyprompt’ also supports R code generation and evaluation, allowing LLMs to run R code. compatible with all LLM providers: Usable with any LLM provider that supports chat completion. Use included LLM providers such as Ollama (local PC/on your own server), OpenAI, OpenRouter (offering various providers, including Anthropic), Mistral, Groq, XAI (Grok), or Google Gemini. Or easily write your own hook for any other LLM provider.
The {tidyprompt} #RStats 📦 *under development* is designed for prompting your #LLMs “the tidy way.” It includes function calling, so you “can define your own R functions and give the LLM the ability to call them.”
By Luka Koning & Tjark Van de Merwe #GenAI
tjarkvandemerwe.github.io/tidyprompt/
Sven
#skyhounds #greyhound #lurcher
The Columbia public health school published this Q&A about my work on causality & algorithmic fairness. Read on to learn how the IMF's imposed austerity policies post-2009 financial crisis led me down a path toward studying causal inference*
*not among the intended effects
shorturl.at/KkeWu
Interesting but not surprising. The next frontier is leveraging this kind of daily activity pattern data to improve health and well-being rather than harm it.
For anyone curious, I've never written a browser extension before, and I only know enough Javascript to be dangerous in Shiny/Quarto. I took a screenshot of a bioRxiv page and pasted into Claude 3.5 Sonnet, and asked it to build one for me. Worked with minimal follow up prompts. ~5 minutes in total.
A screen recording of the RStudio IDE. After highlighting some code, a dialog box pops up with a selection of keywords; after selecting one called "cli", some erroring code is rewritten. Then, after highlighting a whole function and selecting a new keyword "roxygen", roxygen documentation is drafted on-the-fly.
I've been working a new #rstats package called pal🤖
Pals bring a library of LLM prompts into your RStudio/Positron sessions, editing code directly at the press of a keyboard shortcut.
Read more: simonpcouch.github.io/pal/
This seems pretty cool
A bingo card for (bio)stat researchers! I’m dangerously close to a black-out…
What's the statute of limitations on backlogged random thoughts and opinions that didn't get posted at The Other Site? Asking for a friend...
Screenshot of Wikipedia entry on passion fruit
I'm going to blame my dad, who's from South Africa...
I was about to post something to the effect of “in North America we call this a passion fruit” but glad I googled and discovered that granadilla and passion fruit are slightly different #themoreyouknow
Trying again with a working link:
We’ve just posted a new Contract Assistant Professor position at UMN Biostat & Health Data Science! Happy to discuss with anyone interested.
#biostatistics #healthdatascience #academicjobs
hr.myu.umn.edu/jobs/ext/365...
Thanks for the heads up! Didn’t notice since I’m automatically logged in.