Edouard Bonneville

Potentially doi.org/10.1177/0962... ? Otherwise cc @margaritamb.bsky.social , who has very readable papers on DAGs/missing data

Will your favourite open source software tools soon be closed off? Freely available software relies on credit. AI is destroying that.

New post, on how AI is coming for open source software: kucharski.substack.com/p/will-your-...

Your Brain on AI: Is ChatGPT making us mentally lazy? Your Brain on AI: Is ChatGPT making us mentally lazy? page for Normal Curves

cc @kristinsainani.bsky.social and @reginanuzzo.bsky.social , who discuss the statistical limitations of this preprint on their podcast: www.normalcurves.com/your-brain-o...

@lshtm.bsky.social will be running a 3-day online short course on using multiple imputation to handle missing data on 23-25th June 2026. Teaching staff include James Carpenter, Ruth Keogh, Clémence Leyrat, and myself. Further details about the course at www.lshtm.ac.uk/study/course...

INTRODUCTION

Modern-Day Oracles or Bullshit Machines?

Jevin West (@jevinwest.bsky.social) and I have spent the last eight months developing the course on large language models (LLMs) that we think every college freshman needs to take.

thebullshitmachines.com

A "methods primer" article in the journal "BMJ Medicine", titled "Factors associated with: problems of using exploratory multivariable regression to identify causal risk factors"

We wrote an article explaining why you shouldn't put several variables into a regression model and report which are statistically significant - even as exploratory research. bmjmedicine.bmj.com/content/4/1/.... How did we do?

Belgian AI scientists resist the use of AI in academia Several AI scientists have published an open letter calling for a ban on AI use by students.

Belgian AI scientists are advocating *against* the use of AI in academia. “If independent thinking is no longer encouraged at university, where would it?” apache.be/2025/10/24/b...

Multiple Imputation of Missing Covariates When Using the Fine–Gray Model The Fine–Gray model for the subdistribution hazard is commonly used for estimating associations between covariates and competing risks outcomes. When there are missing values in the covariates includ...

Multiple Imputation of Missing Covariates When Using the Fine–Gray Model. Edouard F. Bonneville, Jan Beyersmann, Ruth H. Keogh, Jonathan W. Bartlett, Tim P. Morris, Nicola Polverelli, Liesbeth C. de Wreede, Hein Putter. Statistics in Medicine. onlinelibrary.wiley.com/doi/10.1002/...

Yep as long as you thought to store your simulation results nicely in a tidy / long format it makes things very easy 😅

Looks nice! Suggestion: with the {targets} approach you can have a quarto file at the end of the pipeline to report your results, using e.g. @ellessenne.bsky.social 's excellent {rsimsum} (ellessenne.github.io/rsimsum/) package for performance measures + making nice plots

Hazards Constitute Key Quantities for Analyzing, Interpreting and Understanding Time‐to‐Event Data Censoring makes time-to-event data special and requires customized statistical techniques. Survival and event history analysis therefore builds on hazards as the identifiable quantities in the presen...

Hazards Constitute Key Quantities for Analyzing, Interpreting and Understanding Time-to-Event Data. Jan Beyersmann, Claudia Schmoor, Martin Schumacher. Biometrical Journal. onlinelibrary.wiley.com/doi/10.1002/...

sorry to miss ACIC; enjoy the weird lasagna pizza nerds

A common sentiment is “I can’t share my code because it’s messy and there might be errors”

But if it’s that untrustworthy, why are you publishing the results it generates?

Tariffs xkcd.com/3073

Alas I reviewed one of these recently.. at least it provided an all-timer along the lines of "A scoping review on how measurement error is handled (=inclusion in review if there was measurement error + a correction method used) showed that over x% of papers suffered from measurement error" 🔥

Opinion | The Tech Fantasy That Powers A.I. Is Running on Fumes (Gift Article) A.I. is just what we need in the post-fact era: less research and more predicting what we want to hear.

This piece from @tressiemcphd.bsky.social starts with a bang and just gets better and better. Gift link.

Interesting read, thanks! Would you have any insights as to why methodological journals do not yet make the code sharing mandatory? Quite frustrating as a reviewer..

Here's hoping "(no wizard required)" becomes the new ": a tutorial"

.. at least compared to the matching step before G-computation in each imputed dataset, which does seem to affect the final estimates 😅

Really nice, thanks! While playing around with the code locally, it was interesting (albeit a bit depressing as a missing data enthusiast) how little the imputation approach seemed to matter e.g. unstratified by exposure/random forests/compatible with the model w/ all exposure x confounder terms..

Estimating hypothetical estimands with causal inference and missing data estimators in a diabetes trial case study ABSTRACT. The ICH E9 addendum on estimands in clinical trials provides a framework for precisely defining the treatment effect that is to be estimated, but

'Estimating hypothetical estimands with causal inference and missing data estimators in a diabetes trial case study', led by Camila Olarte Parra, with Rhian Daniel and David Wright, now available open access in Biometrics: doi.org/10.1093/biom...

Performance evaluation of predictive AI models to support medical decisions: Overview and guidance A myriad of measures to illustrate performance of predictive artificial intelligence (AI) models have been proposed in the literature. Selecting appropriate performance measures is essential for predi...

New work in preprint!

"Performance evaluation of predictive AI models to support medical decisions: Overview and guidance".

Under the wings of the STRATOS initiative.

But @maartenvsmeden.bsky.social said it better already 😜

arxiv.org/abs/2412.10288

Create Tidy Data Frames of Marginal Effects for ggplot from Model Outputs Compute marginal effects and adjusted predictions from statistical models and returns the result as tidy data frames. These data frames are ready to use with the ggplot2-package. Effects and predictio...

{ggeffects} version 2.0.0 is out now! Read more about this #rstats package at strengejacke.github.io/ggeffects

What does {ggeffects} do? The 📦 helps you to understand (complex) models and results by calculating and plotting *estimated marginal means* or more generally, *adjusted predictions*. 🧵

Hello everyone! I teach a class for graduate students on methods for reproducible research. 🔧 🔢

I would interested to hear about the barriers you face in implementing reproducible practices in your research (e.g., version control, code review, reproducible documents).

My fav starter packs so far, a thread:

stats: go.bsky.app/Ki7PjpS
stats: go.bsky.app/7TBN5rX
causal inference: go.bsky.app/FdemGAZ
package devs: go.bsky.app/N1569Qh
data peeps: go.bsky.app/8TdEfdK
medical stats: go.bsky.app/ArqEz36
bioinformatics: go.bsky.app/Ha64Gmv
r-ladies: go.bsky.app/Vgxwa2F

Goose chase meme. Top Panel: Suspicious goose asks "biased relative to what?" Bottom Panel: Goose is chasing a man, yelling "Biased relative to what?"

When a study talks about various things that could bias estimates but never explicitly addresses the analysis goal in the first place