Maarten van Smeden(@MaartenvSmeden) 's Twitter Profileg
Maarten van Smeden

@MaartenvSmeden

Statistician • associate prof • statistics/epidemiology/AI methodology, Julius Center @umcutrecht • own views

ID:231008287

linkhttp://www.mvansmeden.net/ calendar_today27-12-2010 09:52:36

20,1K Tweets

41,3K Followers

830 Following

Follow People
Erik van Zwet(@ErikVanZwet) 's Twitter Profile Photo

New OA paper with Lu Tian and rob tibshirani. We show how to use the thousands of trials from the Cochrane database to improve the statistical inference for any given trial. We may expect gains that are roughly equivalent to doubling the sample size!
onlinelibrary.wiley.com/doi/10.1002/si…

account_circle
Richard Riley (R²)(@Richard_D_Riley) 's Twitter Profile Photo

Allons-y!!  Read my TARDIS study in The BMJ 🎄

Paper has 2 hearts
❤️ healthcare provision at Xmas
❤️ 60 years of

Festive Dr Who episodes (proxy for doctor working at Xmas) associated with lower death rates

Doctor Who BBC Disney+ tinyurl.com/882ny2zd

account_circle
Maarten van Smeden(@MaartenvSmeden) 's Twitter Profile Photo

Your data may be missing
Your estimator may be biased
Your measurements may have errors
Your research question may be undefined
Your causal association may have residual confounding

Please don't worry

Your paper will be published

account_circle
Maarten van Smeden(@MaartenvSmeden) 's Twitter Profile Photo

Machine learners: MACHINE LEARNING OUTPERFORMS STATISTICS when I use statistical models that make assumptions that I know cannot be true

Statisticians: STATISTICS OUTPERFORMS MACHINE LEARNING when I use some random off the shelf machine learning approach with no tuning

account_circle
Maarten van Smeden(@MaartenvSmeden) 's Twitter Profile Photo

absence of evidence ≠ evidence of absence
measurement error ≠ bias towards null
clinical trial ≠ study with patients
multivariate ≠ multivariable
95%CI ≠ 95% probability
p-value ≠ probability H0
correlation ≠ causation
significant ≠ relevant
bias ≠ error
odds ≠ risk

account_circle
Andrew Althouse(@ADAlthousePhD) 's Twitter Profile Photo

Enjoyed this Tim Morris post about “re-randomization” designs that allow participants to enroll multiple times

tpmorris.substack.com/p/intuitive-un…

account_circle
Oisín Ryan(@Oisin_Ryan_) 's Twitter Profile Photo

Want to learn how observational data can yield insights into causal effects? How to specify a target trial & how to use prediction models for causal insight?

Check out our (Wouter van Amsterdam & BP de Vries) new summer school course, August 5 -9 in Utrecht:
utrechtsummerschool.nl/courses/health…

Want to learn how observational data can yield insights into causal effects? How to specify a target trial & how to use prediction models for causal insight? Check out our (@WvanAmsterdam & BP de Vries) new summer school course, August 5 -9 in Utrecht: utrechtsummerschool.nl/courses/health…
account_circle
Wouter van Amsterdam(@WvanAmsterdam) 's Twitter Profile Photo

very excited to have this pre-print online!

some prediction models have good AUC pre- and post-deployment, but are (silently) harming patients when used for treatment decisions

I'll present this at ML4H Dec 10th and will attend after

very excited to have this pre-print online! some prediction models have good AUC pre- and post-deployment, but are (silently) harming patients when used for treatment decisions I'll present this at @SymposiumML4H Dec 10th and will attend #NeurIPS2023 after
account_circle
Erkan Kalafat, MD MSc(@kalafaterkan_md) 's Twitter Profile Photo

Maarten van Smeden Asked Bard to rate the sarcasm level.

Response: The comment is highly sarcastic, reaching 95 out of 100 on the sarcasm scale. Here's why:

g.co/bard/share/7a1…

account_circle
Richard Riley (R²)(@Richard_D_Riley) 's Twitter Profile Photo

📽️NEW VIDEO 📽️

'How to get your article rejected by the BMJ:
12 common statistical issues'

I discuss common stats issues we encounter at The BMJ

Please check before you submit next article 🫣

Santa would check the list twice 🎅

Hope helpful 🙏
youtube.com/watch?v=iu4VsE…

account_circle
Maarten van Smeden(@MaartenvSmeden) 's Twitter Profile Photo

Really proud of this team. Learn about the state-of-the-art causal inference methods this summer in Utrecht from experts that happen to be also excellent teachers
Utrecht University UMC Utrecht

account_circle
Richard McElreath 🦔(@rlmcelreath) 's Twitter Profile Photo

Maarten van Smeden going add these to CV under special skills:
• Making research slower
• Reducing research intelligibility
• Inflating research budgets
• Rendering results less impressive
• Boring

account_circle