Gordon Forbes

NIHR Doctoral Fellow at King's College London Biostatistics Unit

[Marion Campbell @MarionKCampbell]

A major update to the CONSORT reporting guidance for clinical trials was published last week.
📌 CONSORT 2025 replaces all previous versions and should be used from now on.
So what’s new and what’s different? 1/7
#MethodologyMonday #116
(COI - I am a co-author)

[Gael Varoquaux 🦋 @GaelVaroquaux]

#ICLR2025 @MarineLeMorvan presents "Imputation for prediction: beware of diminishing returns": poster Thu 24th

Concludes 6 years of research on prediction with missing values: Imputation is useful but improvements are expensive, while better learners yield easier gains.

[Laure Wynants @laurewynants.bsky.social @laure_wynants]

What a great day yesterday! 🎉 Dr. Shan Gao and Dr. Elena Albu @sibipx successfully defended their PhDs on the dynamic prediction of CLABSI using EHR data. A huge milestone after four years of dedicated research!

[Gordon Forbes @GordonJForbes]

Grade A trash talking

[Dr Kareem Carr @kareem_carr]

Most people don't know this but AI is just statistics. It's a special kind of statistics where you can use data without permission, spend billions of dollars of resources on a single model, and you aren't personally responsible when it fails in idiotic ways and/or hurts people.

[Bia @BeatrizGietner]

No, we are not moving from overleaf to quarto

[Andrew Vickers @VickersBiostats]

Why do we do external validation of prediction models? Why do we assess calibration? What should we do if a model is miscalibrated? Should we be doing "local recalibration" so that models work for our patients? https://t.co/aJDJi7EuhV

[Gordon Forbes @GordonJForbes]

@BenVanCalster It should be the only research funded until everyone else sorts out their act.

[Gordon Forbes @GordonJForbes]

Imagine getting your stats grant pulled fro investigating bias in estimators or the discrimination of a prediction models.

[Rachel Cohen Booth @rcobooth]

Per @darbysaxbe, from a program officer at the National Science Foundation, federal grants can now be pulled for containing any of these keywords. Absurd and dangerous

[Rachel Cohen Booth @rcobooth]

Per @darbysaxbe, from a program officer at the National Science Foundation, federal grants can now be pulled for containing any of these keywords. Absurd and dangerous

[Gordon Forbes @GordonJForbes]

Interesting to find that the industry does more interim analysis.

I feel like the industry is more motivated + able to stop trials and reallocate resources than academia, where the early ending of a grant could lead to difficulties with funding fixed-term contracts.

[Steven Julious @StevenJulious]

Justed had paper published with @qiangzh83769692 , @mdimairo , Jen Lewis and Zihang Yu entitled

"Reporting and communication of sample size calculations in adaptive clinical trials: a review of trial protocols and grant applications"

https://t.co/oXpiJ9PSv5

[Steven Julious @StevenJulious]

Justed had paper published with @qiangzh83769692 , @mdimairo , Jen Lewis and Zihang Yu entitled

"Reporting and communication of sample size calculations in adaptive clinical trials: a review of trial protocols and grant applications"

https://t.co/oXpiJ9PSv5

[Gael Varoquaux 🦋 @GaelVaroquaux]

I tried to give my high-level learnings of 15 years of machine learning for health

[Gordon Forbes @GordonJForbes]

@naked_statist transposed conditional fallacy

[Dr Kareem Carr @kareem_carr]

still can't get over the fact that people in tech say inference when they mean prediction. we already have a word for prediction. it's called prediction.

[Asjad Naqvi @AsjadNaqvi]

Dropping a new @Stata guide where we learn how to fine tune #graphs for academic #publishing. Here we talk about using a very small space for graphs & fine tuning it w label sizes, line colors & widths, legend placement, use of decimals, axis scaling etc.

https://t.co/7LKKYxTKeW

[Gordon Forbes @GordonJForbes]

Simulation studies just got more complicated than x <- rnorm(100).

[Matthew Berman @MatthewBerman]

.@Microsoft just dropped TinyTroupe!

Described as "an experimental Python library that allows the simulation of people with specific personalities, interests, and goals."

These agents can listen, reply back, and go about their lives in simulated TinyWorld environments.

[Gordon Forbes @GordonJForbes]

Come and work with us! ���� We're seeking a postdoc to help develop an R package to estimate sample sizes for machine learning. Programming, simulation, longitudinal or ML skills = 🌟
👉 Apply https://t.co/iGtycS36Ar
🗓️ Deadline 2/12
💬 Questions? Get in touch!

#rstats #stats

[Gordon Forbes @GordonJForbes]

Evidence that waitlist control trials lead to biased effect estimates in psychotherapy trials for depression.

Caution needed as findings show substantial heterogeneity and the wait list control trials differ from other trials on key characteristics.