@f2harrell.bsky.social you were right. Coverage for TMLE xgboost grid search size 5 appeared better than 20 but still the coverage is disappointing. I’ve updated this examination for my learning. Thanks again for the guidance. www.kenkoonwong.com/blog/tmle/

agree with correct covariates but DGP incorrectly specified. My sim didn't have iv, mediators or colliders. but it'd be interesting to include these on the examination. I assume BS is bootstrap, not sure what EF is. in TMLE we can use influence function to calculate SE too.

Will do! Thanks for the pointers. Prelim result on grid search size of 5 so far is proving the disappointment right.

Kudos! Well deserved!

:) that’s when cloud computing was helpful. I’ll re run it to assess coverage and report back hopefully next week.

@josephmarcusid.medsky.social reminds me of your conjunctivitis investigation.

Yes I have. They’re great! I made several references to her blogs. I just wanted to test the doubly robustness myself and see if bias and variance really gets minimized in TMLE using tree based method. Her blog and cheat sheet gave clear instructions to build TMLE from ground up with using package

Mind you, this is like a total revamp of your powerpoint slides :(

1. Download quarto.org . If you already have Rstudio, it should already have quarto
2. Create quarto presentation (basically just a markdown)
3. Make your presentation as a markdown and paste the iframe code given.
4. render as html

Final product may look something like this

Nothing straightforward to embed iframe that I know of unless you’re using html rendered slides from markdown or rmarkdown. Even if u reproduce the interactive plot you still need some form of platform that can use js

Thanks. That was included :) the query will pick up any posts that has idweek, with or without hashtag.

D. All of the above!

And @sairabt.bsky.social and @josephmarcusid.medsky.social

@jonathanrydermd.bsky.social @josephmarcusid.medsky.social

This is genius!!!

Thanks to @josephmarcusid.medsky.social and @jonathanrydermd.bsky.social for the motivation and insight!