Lightnews — Scholar-powered news

Denis Mongin

@denismongin.bsky.social

530 followers 550 following 810 posts

Started with lasers and nonlinear optics, now with biomedical informatics, data and statistics in health sciences. Research in #Metaresearch, #rheumatology and #publichealth at HUG and University of Geneva. #Rstats ninja, #Openscience

Posts Media Videos Starter Packs

Pinned

Denis Mongin @denismongin.bsky.social · Nov 26

Do people register their #randomizedtrials before starting it ?
We studied more than 1000 RCT in #rheumatology to answer this question, to see how it evolved in time and how it varied between RCTs⬇️

Just published in @jclinepi.bsky.social :
jclinepi.com/article/S089...

#Metaresearch #EpiSky

Prospective registration of trials: where we are, why, and how we could get better

Transparent trial conduct requires prospective registration of a randomized controlled clinical trial (RCT) before the enrollment of the first participant. We aimed to (1) estimate the proportion of R...

jclinepi.com

2 9 21

Denis Mongin @denismongin.bsky.social · 1h

How does you handle this kind of situation for statistics if you are not trained in stat ?

bsky.app/profile/rand...

Randy Au @randyau.com · 5h

lol, I was toying with gemini to explain a bunch of LLM transformer background knowledge to me since it's way out of my self study depth and for no good reason at all it pulls a diagram of a literal electronics transformer at the end

screenshot showing a large image showing of an electronics transformer works including two wire loops, transformer core, AC power on a primary circuit and lightbulb on the primary circuit, with variables for power and voltage turns, cretited to getty images. Immediately above in text are mathematical formulae and a discussion of the billions of parameters in an LLM transformer model

Denis Mongin @denismongin.bsky.social · 2h

Hi wahou me neither

Reposted by Denis Mongin

econmaett @econmaett.github.io · 5h

I didn't know you could write nested loops in R this way
#rstats
www.quantstart.com/articles/Aut...

for (p in 1:4) for (d in 0:1) for (q in 1:4)

5 3 28

Reposted by Denis Mongin

Andrew Heiss @andrew.heiss.phd · 18h

I really really really liked The Oatmeal’s take on AI here theoatmeal.com/comics/ai_art

A cartoonist's review of AI art - The Oatmeal

This is a comic about AI art.

theoatmeal.com

3 23 67

Denis Mongin @denismongin.bsky.social · 1d

It remembers me of this book, having the same kind of thought

www.goodreads.com/book/show/29...

Goodreads

Discover and share books you love on Goodreads.

www.goodreads.com

Reposted by Denis Mongin

HydrologyNL @hydrology.nl · 2d

The Koyukuk River near Huslia, Alaska. Final version for today.

A palette of 6 colourful panels, i.e. six maps of the same area with 6 different colour schemes.

2 8 26

Denis Mongin @denismongin.bsky.social · 4d

m.youtube.com/watch?v=BkXi...

Marre de ce cinéma français - La chronique d'Akim Omiri dans "La dernière"

YouTube video by Radio Nova

m.youtube.com

Denis Mongin @denismongin.bsky.social · 4d

Il s'est passé quoi en 2019 ?

Reposted by Denis Mongin

ProPublica @propublica.org · 5d

Revlimid has its origin in a pill that cost patients $7.50 each.

Decades later, the cancer drug costs more than $18,000 for a month’s supply — even though it still only costs about 25 cents to manufacture.

Read @davidarmstrongx.bsky.social's award-winning investigation:

The Price of Remission: This Cancer Drug Saves Lives — but Costs a Fortune. I Wanted to Know Why.

When I was diagnosed with cancer, I set out to understand why a single pill of Revlimid cost the same as a new iPhone. I’ve covered high drug prices as a reporter for years. What I discovered shocked ...

www.propublica.org

48 660 1.2K

Reposted by Denis Mongin

Jeff Lockhart @jwlockhart.bsky.social · 5d

I'm excited to share my new paper with a former student, Tommy Smith. "'This Work Would Not Have Been Possible without...': The Length of Acknowledgments in Sociology Books"

Open Access in @sociusjournal.bsky.social : doi.org/10.1177/2378...

Main figure from the paper. Violin plots showing the length of acknowledgments sections in books by sociologists, broken out by gender, race, sexuality, parents’ education, millennium of author’s PhD, and publisher type.

13 72 210

Reposted by Denis Mongin

Sharon Machlis @smachlis.bsky.social · 7d

Just learned: It's so easy to sort strings that are numbers in proper numerical order with the {stringr} #Rstats 📦's `str_sort()` function and numeric = TRUE!!
x <- c("8", "10", "1", "40")
str_sort(x, numeric = TRUE)
[1] "1" "8" "10" "40"

6 11 73

Denis Mongin @denismongin.bsky.social · 7d

Hahaha
Exactly that

Denis Mongin @denismongin.bsky.social · 7d

Wait until people use this to analyse LLM generated data

bsky.app/profile/o.si...

Olivier Simard-Casanova @o.simardcasanova.net · 8d

I cannot believe that Microsoft actually branded this as “vibe working”

Microsoft launches ‘vibe working’ in Excel and Word

Vibe working is all about Office’s new Agent Mode.

www.theverge.com

1 1

Denis Mongin @denismongin.bsky.social · 7d

And then you use excel "vibe coding" fonction to analyse the data.
And chatgpt to write the paper.

Denis Mongin @denismongin.bsky.social · 8d

The author's thread about this preprint

bsky.app/profile/jami...

Jamie Cummins @jamiecummins.bsky.social · 21d

Can large language models stand in for human participants?
Many social scientists seem to think so, and are already using "silicon samples" in research.

One problem: depending on the analytic decisions made, you can basically get these samples to show any effect you want.

THREAD 🧵

The threat of analytic flexibility in using large language models to simulate human data: A call to attention

Social scientists are now using large language models to create "silicon samples" - synthetic datasets intended to stand in for human respondents, aimed at revolutionising human subjects research. How...

arxiv.org

1 1 4

Denis Mongin @denismongin.bsky.social · 8d

Well, at home...
4.5 was 366B, so difficult to run on cheap consumer GPU

Denis Mongin @denismongin.bsky.social · 8d

Dear Biostat fellows, fear the excel vibe coding wave coming

Olivier Simard-Casanova @o.simardcasanova.net · 8d

I cannot believe that Microsoft actually branded this as “vibe working”

Microsoft launches ‘vibe working’ in Excel and Word

Vibe working is all about Office’s new Agent Mode.

www.theverge.com

Reposted by Denis Mongin

Vincent Arel-Bundock @vincentab.bsky.social · 9d

{tinytable} 0.14.0 for #RStats makes it super easy to draw tables in html, tex, docx, typ, md & png.

There are only a few functions to learn, but don't be fooled! Small 📦s can still be powerful.

Check out the new gallery page for fun case studies.

vincentarelbundock.github.io/tinytable/vi...

1 39 130

Reposted by Denis Mongin

Tim Kellogg @timkellogg.me · 11d

please wait like a day or two, i have no real process for remembering who the good accounts are

go.bsky.app/LFAZcGE

3 4 44

Denis Mongin @denismongin.bsky.social · 10d

The diff between unadjusted and adjusted give an indication about the amplitude of the confounding

Denis Mongin @denismongin.bsky.social · 11d

Marche aussi pour ceux qui beuglent pour la liberté d'expression, qui sont les premiers à ma museler

Denis Mongin @denismongin.bsky.social · 13d

J'espère que le plupart des gens trouvent au contraire cela rassurant pour l'état de droit

chapeau d'un article du monde:

"En direct, procès Sarkozy-Kadhafi : l’ex-président, condamné à cinq ans de prison avec mandat de dépôt à effet différé, dénonce un jugement « d’une gravité extrême pour l’Etat de droit »
« Cette injustice est un scandale », a réagi l’ex-chef de l’Etat, reconnu coupable d’association de malfaiteurs dans l’affaire du financement libyen. Le tribunal judiciaire de Paris évoque des « faits d’une gravité exceptionnelle » pour justifier l’exécution provisoire de la peine. Cela veut dire que M. Sarkozy n’échappera pas à la prison, même en ayant fait appel."

Denis Mongin @denismongin.bsky.social · 13d

Who?

Reposted by Denis Mongin

Prem Thakker ツ @premthakker.bsky.social · 16d

Trump's Executive Order on Antifa is written such that someone recording masked agents snatching people off the streets, or asking these agents what they're doing, can be deemed a "terrorist."

Section 1. Antifa as a Terrorist Threat. Antifa is a militarist, anarchist enterprise that explicitly calls for the overthrow of the United States Government, law enforcement authorities, and our system of law. It uses illegal means to organize and execute a campaign of violence and terrorism nationwide to accomplish these goals. This campaign involves coordinated efforts to obstruct enforcement of Federal laws through armed standoffs with law enforcement, organized riots, violent assaults on Immigration and Customs Enforcement and other law enforcement officers, and routine doxing of and other threats against political figures and activists. Antifa recruits, trains, and radicalizes young Americans to engage in this violence and suppression of political activity, then employs elaborate means and mechanisms to shield the identities of its operatives, conceal its funding sources and operations in an effort to frustrate law enforcement, and recruit additional members. Individuals associated with and acting on behalf of Antifa further coordinate with other organizations and entities for the purpose of spreading, fomenting, and advancing political violence and suppressing lawful political speech. This organized effort designed to achieve policy objectives by coercion and intimidation is domestic terrorism.

68 340 800

Denis Mongin @denismongin.bsky.social · 16d

Gut feeling, I would tend to maximize the available info to impute, so full dataset then subset