iiinova — where your digital self keeps evolving

Joseph Lemaitre @josephlemaitre.bsky.social · 22/01/2026

Really neat bit on Rt confidence (also loving the blog write-up, more edible than skimming the full paper)

Joseph Lemaitre @josephlemaitre.bsky.social · 14/11/2025

NHSN and other epi datasources back again. I imagine there is quite some work under the scene, thanks CDC folks we've been missing you.

010

Joseph Lemaitre @josephlemaitre.bsky.social · 22/10/2025

Inconsistent is what I dislike the most

010

Reposted by Joseph Lemaitre

Sam Abbott @seabbs.bsky.social · 17/10/2025

#epinowcast 0.4.0 is out! This release has been a long time coming and contains work from many contributors. It includdes new features, better better and clearer documentation. package.epinowcast.org/news/index.h...

package.epinowcast.org

Changelog

1107

Joseph Lemaitre @josephlemaitre.bsky.social · 19/08/2025

last time I tried it the system library autocomplete was super slow... not sure if I had a poor config, but I'd say half a second to get the proposal. I was really neat though

010

Joseph Lemaitre @josephlemaitre.bsky.social · 12/08/2025

just learnt that vs code recently added native git blame decorations with setting: "git.blame.editorDecoration.enabled": true good bye complex extensions, goodbye gitlens 👋

010

Joseph Lemaitre @josephlemaitre.bsky.social · 24/07/2025

Great read. If priced out of a bet, the best things is to sell the opportunity to bet to a higher bankroll party (if possible). There are many cases in life where uncertainty can be traded at discount for certainty (insurance policy is a bet on a home getting flooded, bank loans, ...)

010

Joseph Lemaitre @josephlemaitre.bsky.social · 17/07/2025

I don’t know R but these things are not plain text files?

100

Joseph Lemaitre @josephlemaitre.bsky.social · 15/07/2025

I've been sneaking on that course (by Sam Abbott, Thomas Robacker, Nick Reich) for the past few days on github (did not see the deployed version) and it is really good learning material for learning epi forecasting: nfidd.github.io/sismid/

nfidd.github.io

Nowcasting and forecasting infectious disease dynamics - SISMID – NFIDD SISMID

Nowcasting and forecasting infectious disease dynamics - SISMID

120

Joseph Lemaitre @josephlemaitre.bsky.social · 11/07/2025

James does really great science ! This is a great job

020

Joseph Lemaitre @josephlemaitre.bsky.social · 11/07/2025

This project is supported by the Swiss National Science Foundation @snsf.ch, on a project grant specifically written for this reproducibility project! It’s a great win 🎉

100

Reposted by Joseph Lemaitre

Dorothy Bishop @deevybee.bsky.social · 10/07/2025

Wow. This is an amazing project with a huge amount of work behind it. Even if you aren't into drosophila, it's worth a look

04114

Joseph Lemaitre @josephlemaitre.bsky.social · 10/07/2025

.. but found Aider (usually Sonnet Architect + DeepSeek implementor) and Claude Code extremely productive. Task is also important, it is incredible and safe for plotting and front-end... found it slows me down on many pure modeling task. All anecdotal obviously.

020

Joseph Lemaitre @josephlemaitre.bsky.social · 10/07/2025

yes, very nice someone studied it so well. I am very curious on their hinted follow-up on agent. I always fount LLM clunky, and cursor frustrating (there is some kind of bad context compression or something) ...

110

Reposted by Joseph Lemaitre

Mark A. Hanson @hansonmark.bsky.social · 10/07/2025

But in fact, fly immunity is highly reproducible! 🎊 ~80% of claims could be verified. Moreover, some challenged claims just reflect the field advancing its knowledge/tools. The lesson: if the tools are good and the research largely exempt from direct translation pressures, science works. 3.5/n

A chart depicting the track of claims into verified, unchallenged, challenged, partially verified, or mixed.

Figure 1: Flow of the study.
Claims were first classified in five categories based on literature. 45 unchallenged claims were selected to be experimentally tested and
classified either as challenged or as verified. We further classify some unchallenged as consistent or inconsistent depending on their
consistency with current knowledge.

24918

Reposted by Joseph Lemaitre

Mark A. Hanson @hansonmark.bsky.social · 10/07/2025

When everything is considered together (w/ caveats), the most prestigious institutes are the most likely to produce irreproducible work. Journal tier was a more minor effect, and in fact, "high-impact" but not "trophy" journals publish the most reproducible science. But, many caveats (cont.) 4.3/n

A chart showing various factors and their odds ratios and relative contributions to irreproducibility in a multivariate mixed model.

Figure 10. Multivariable analysis of predictors of claim irreproducibility.
The forest plot reports the odds ratio and the 94% highest density interval of each covariate effect on
a claim becoming challenged.

12410

Reposted by Joseph Lemaitre

Richard Sever @richardsever.bsky.social · 02/07/2025

more than that but right order of magnitude. I always say $10^1 for preprint; $10^3 for journal article; $10^5 for research. most costs are staff not tech because of responsible screening, customer service, etc.

2133

Joseph Lemaitre @josephlemaitre.bsky.social · 01/07/2025

Seems like it cannot do these little neat interactive javascript diagrams but claude has gotten better at ascii

000

Joseph Lemaitre @josephlemaitre.bsky.social · 01/07/2025

I remember you talking about this :D This is great (I get a Error though, only when I am logged with Claude). Hope you get back the you get a list of the questions

100

Joseph Lemaitre @josephlemaitre.bsky.social · 30/06/2025

I'm not in epiengage but I'll try with something more unformed ! Squares with LLMs ability in green field projects vs already existing stuff.

110

Joseph Lemaitre @josephlemaitre.bsky.social · 30/06/2025

Oh no, I now have to test twice the hypothesis 😦

020

Joseph Lemaitre @josephlemaitre.bsky.social · 30/06/2025

interesting, somehow never managed to use LLM for this, they always compromise my ideas/go too far.... I find them useful for nearly everything else (finding flaws in ideas, developing the concept). Perhaps my notes are too scattered. Or my prompt is bad.

100

Joseph Lemaitre @josephlemaitre.bsky.social · 30/06/2025

(Obvious disclaimer that LLM use comes with many shortcomings, and world model is something that is hard to define. Just hope for more nuance.) arxiv.org/pdf/2403.154... is also a good read (yay for independent researchers).

arxiv.org

000

Joseph Lemaitre @josephlemaitre.bsky.social · 30/06/2025

Obviously, this last bit also highlights that the world-model is relatively brittle. But still, it exists. That's for non-coders. For coders, there is a before and after using Claude Code. A lot of the academic literature on LLMs brushes over code (like the big PNAS piece recently forgot who).

100

Joseph Lemaitre @josephlemaitre.bsky.social · 30/06/2025

Chess is even my pet example to demonstrate how world-models are built by Transformers + Data. The proportion of legal moves is very low, and a LLM always plays legal moves and quite good ones. Gary's claim is caused by the chat environment that destroys the context windows for chess-playing.

100

Joseph Lemaitre @josephlemaitre.bsky.social · 30/06/2025

This opinion (LLMs have no world model) is repeated at every AI track in IDD conferences, and while it's true for a narrow definition of world model, I am sure it does not square with audience experience. Contrary to what Gary says, LLMs play good (1'800 ELO) chess: dynomight.net/more-chess/.

110

Joseph Lemaitre @josephlemaitre.bsky.social · 05/06/2025

Slightly related to what you describe at the IDD afternoon and your comment about complexity

110

Joseph Lemaitre @josephlemaitre.bsky.social · 04/06/2025

This is a good read ! Paul I think you'd enjoy this paper as well arxiv.org/pdf/2503.02113. You touch that in the conclusion but something is to be written in a course on how to connect simple NN one understands with the magic that happens when you add so many parameters.

arxiv.org

120

Reposted by Joseph Lemaitre

Sam Abbott @seabbs.bsky.social · 07/05/2025

Which academic communities have a good record of sharing domain-specific open source tools? Bioinformatics and geospatial analysis comes to mind. I am trying to think about if they are examples to follow or if there is some difference in the domain (to Infectious disease modelling).

5114

Joseph Lemaitre @josephlemaitre.bsky.social · 29/05/2025

xkcd.com/2347/ this is true for academia and the question is how to identify these projects. For software, systematically scanning repository for dependencies perhaps...

xkcd.com

Dependency

110

Reposted by Joseph Lemaitre

Michael "Shapes Dude" Betancourt @betanalpha.bsky.social · 28/05/2025

Earlier this month I gave a workshop on Bayesian methods in epidemiology. With my hosts I developed a pretty elaborate, demonstrative analysis for the workshop which I am happy to share publicly here. HTML: betanalpha.github.io/assets/chapt... PDF: betanalpha.github.io/assets/chapt...

betanalpha.github.io

Bayes-by Shower

1104

Reposted by Joseph Lemaitre

Sam Abbott @seabbs.bsky.social · 29/01/2025

Note that one of these projects is about delay estimation (described by one 5 year old as double extra boring) but critical to get right for most/all downstream modelling use cases.

372

Reposted by Joseph Lemaitre

Sam Abbott @seabbs.bsky.social · 22/01/2025

For a few years, I have been paying out of pocket for the epinowcast forum as I thought it was important to have a venue for people to collab. There is some great stuff kicking off at the moment which is lovely to see. community.epinowcast.org/t/include-a-...

community.epinowcast.org

Include a simple reference model

Baselines considered: no correction baseline (naïve but not realistically what epi colleagues would expect) cut off last few weeks and repeat the previous week(s) (i.e. thinking that the data will b...

183

Joseph Lemaitre @josephlemaitre.bsky.social · 17/01/2025

This is a very interesting talk by @scarpino.bsky.social continuing the exploration of the predictability of an influenza season using fundamental and practical methods www.youtube.com/live/8RslYIn...

youtube.com

YouTube

Share your videos with friends, family, and the world

141

Joseph Lemaitre @josephlemaitre.bsky.social · 07/01/2025

This includes understanding how these methods can improve inference and prediction through new methodologies and by unlocking access to underused data sets.

000

Joseph Lemaitre @josephlemaitre.bsky.social · 07/01/2025

The goal is to highlight the role that artificial intelligence methods (here in its broader sense, comprising machine learning and other recent advances in statistical and computational methods) play in the study of infectious disease dynamics.

100

Joseph Lemaitre @josephlemaitre.bsky.social · 07/01/2025

The deadline for the Epidemics special issue on Artificial Intelligence for Infectious Disease Dynamics has been extended to 30 April 2025, so there is still time to submit your paper 📝 Feel free to contact me or another guest editor for more information 😊 www.sciencedirect.com/journal/epid...

122

Joseph Lemaitre @josephlemaitre.bsky.social · 15/12/2024

The most generic citations in the main (Disease causes XX death [NEJM 2017] and is seasonal [PNAS 1992]) while the actual building blocks that made the research possible buried in SI… But i always found SI concept to be strange, as I always need to read it every time/more informative than main.

010

Joseph Lemaitre @josephlemaitre.bsky.social · 06/12/2024

The studies informing stratified parameters in IDD models often have partitions (e.g age group) different from the model partitions. Most papers takes the median/mean estimates, but a correct solution exists: see this great paper/R pkg (C. Pearson, @lucygoodfellow.bsky.social @mert0248.bsky.social)

071

Joseph Lemaitre @josephlemaitre.bsky.social · 19/11/2024

Very interesting write up on OWD pandemic work. In general, it’s always surprising how small most teams doing important thing are.

010

Joseph Lemaitre @josephlemaitre.bsky.social · 14/11/2024

Just discovered this excellent book on genomics from Tanja Stadler et al. Free pdf offered here: decodinggenomes.org. The chapters I've read hit a biology/statistics sweet spot for me and probably other IDD researchers.

075