Glad to share our full paper on why deployed clinical prediction models should not discard predicted risks in favour of thresholds with @rexwp.bsky.social and @aidybarnett.bsky.social now up on @jclinepi.bsky.social. A brief explainer (1/4):
#statsky #rstats
authors.elsevier.com/a/1m1D83BcJQ...
😂😂 love this engagement
I reckon this would be particularly helpful for teams that want to work on projects and have a defined structure to reduce cognitive overhead when switching between projects or reviewing code. Hopefully you do too!
The result is that I can run a few lines of code to get a new project started and on github with all my usual adjustments from the standard project setup that targets::init() gives you.
There are a few at the moment that I frequently use: one for a basic targets template, and a couple specific to forecasting. The forecasting template is on a private repo but the basic targets pipeline isn't (github.com/adha-rex/tar...).
Leaning a bit on how I would do the same thing with internal R packages, I now have released versions on GitHub for project templates for a given purpose and this allows me to pull the latest version and track that within projects.
At my work, we are often delivering on many projects with pretty short timelines. I still like to code in R and have a nice structure to my project, no matter it's size, so I've spent a bit of time streamlining how to get these projects started quickly. Here's my trick...
I had this (very surprising) issue before and my solution was to use all.equal() instead.
github.com/ropensci/pre...
Has to beat your interest rate plus your interest rate * your marginal tax rate (if your investment is income producing or if there aren't some other tax breaks associated with it).
Liking the positive vibes of BlueSky so far but finding it tricky to find my 🇦🇺 stats peeps.
Thought I'd try a starter pack to help us find each other. Here's who I've found so far.
Let me know if you'd like to be added #StatsSky
go.bsky.app/QGgfiQV
New paper on falls risk prediction led by @rexwp.bsky.social. Predicting falls risk in hospital tends to be time-intensive for clinicians, so we wanted to see whether we could achieve similar predictive performance with routine EMR data. Using Cox models actually led to BETTER discrimination (1/2)
Since absolutely nobody asked for it, I went ahead and made a package for the MyHospitals API: rwparsons.github.io/readaihw/
Very keen to be responsive to feedback and build out features which users will find helpful so please give it a go and let me know what you would like to see!