Pierre Bauvin (@BauvinPierre) profile picture

Data Scientist, phD 🔎 #liverdisease #obesity #preventivehealth DAG-advocate, causality-passionate, walking the two worlds R & Python, biostats & ML

Joined August 2014
Photo by Pierre Bauvin (@BauvinPierre)Photo by Pierre Bauvin (@BauvinPierre)Photo by Pierre Bauvin (@BauvinPierre)
Replying to @Apple
@Apple say their new sleep apnea prediction tool for Apple Watch is "validated in a clinical study — unprecedented in size for sleep apnea technology" - yet no reference provided Concerning, as 'validated' does not equal valid Can anyone source the study #AppleWatch10
1
2
6
717
🧑‍🎓✍️⚕️🤖Attention PhD researchers searching a postdoc. Do you want to work on cutting edge research in causal AI? Help us bridge healthcare and AI. Join us and help us shape the research of the national chai.ac.uk hub. jobs.ac.uk/job/DHO645/resear… #research #postdoc
11
1
18
3,572
Pierre Bauvin retweeted
Mark your calendar for PyData Paris 2024, Sep 25-26 at Cité des Sciences @citedessciences. With four amazing keynote speakers: Arthur Mensch @arthurmensch Katharine Jarmul @kjam, Olivier Grisel @ogrisel, Merve Noyan @mervenoyann. Check out our blog post: medium.com/@PyDataParis/anno…
3
43
3
93
38,768
📣Our 🆕 paper Causal Inference is Not Just a Statistics Problem is out! @malco_barrett, @travisgerke, and I show that you can have 4 data sets with identical summary stats & visuals but very different data generating mechanisms-statistics alone can't tell you what to adjust for!
20
328
15
1,306
219,753
Pierre Bauvin retweeted
5
91
6
1,098
70,960
Just launched with @IavorBojinov a new column Causal inference for everyone @TheHDSR @harvard_data. We thank our "muse" and editor-in-chief @XiaoLiMeng1 hdsr.mitpress.mit.edu/pub/la…
11
2
39
5,967
Seeking a Research Scientist or Postdoctoral Researcher at Causal Inference Team (W23124) | RIKEN riken.jp/en/careers/research…
9
8
6,299
Sounds like it was all a panic manufactured by the press. But doctors with hundreds of thousands of followers here are more than happy to propagate misleading info before hearing all the details. ukbiobank.ac.uk/learn-more-a…
Scaremongering abt a situation where individual data was probably not given out (so companies cannot ID the individual people) gets you a lot of likes on twitter, but contributes to an anti-progress, safetyist culture. The article is super vague, there is *some data* that was shared & we do not know why or how. If it's done with anonymisation, why is this an issue? I have witnessed how transformative data from the UK BioBank has been to research. Why are there so many people hellbent on undermining progress for a bit of online attention?
1
2
1
4
2,829
Pierre Bauvin retweeted
📢NEW PAPER A step away from my methods work to investigate #clinicalpredictionmodels implemented within the #NHS, and their potential clinical benefit when combined with a #patientcentred intervention specialised in care coordination, patient #wellbeing and #empowerment 🌟
This Post is from an account that no longer exists. Learn more
1
4
18
4,140
Pierre Bauvin retweeted
Chapter Two is here! 🍿 voltrondata.com/codex/standa…
Awesome initial post on how to design, build, and maintain composable data systems from @VoltronData. voltrondata.com/codex/a-new-…
2
2
1,458
Pierre Bauvin retweeted
I'm intending the platform to be used by any field that studies humans (e.g. epidemiology and demography as well). Each link is driven by one or more causal claims in the literature (you can see the citations driving the link by clicking on it)
Replying to @jon_mellon
I am not a social scientist but I am very much curious to know how you initially designed that causal network platform. A consensus among colleagues? A personal or published review of the literature?
2
2
1
11
5,589
Pierre Bauvin retweeted
🚨We have a new working paper on AI & work🚨 In pre-registered experiments at BCG, the elite consulting firm, consultants using the GPT-4 AI finished 12.2% more tasks, completed tasks 25.1% more quickly & produced 40% higher quality results. Big gains. 1/ oneusefulthing.org/p/centaur…
48
562
163
2,477
985,305
Very glad to see our #MachineLearning weight loss prediction model in the Lancet Digital Health : authors.elsevier.com/sd/arti… Great collaboration between @univ_lille, @Inria and #IMISOPHIA A quick methodological feedback from my point of view, as a co-first author (1/3)
1
1
2
142
Using ML in healthcare is a rewarding challenge. Beyond the innovative method, the main difficulty lies in 1) collaboration between scientists who sometimes speak different languages 2) achieving the balance between best performance and best interpretability of the model (2/3)
1
20
The devil is in the detail: the ML model and the choice of variables must be sufficiently transparent to maintain the confidence of practitioners and patients alike, not to mention the need for a restricted set of variables for sake of validation in external cohorts! (3/3)
17