christosargyrop.bsky.social
Division Chief, Nephrology UNMHSC
(im)personal views & account #zerokale
Firm believer : caveat emptor & stupidity is immoral
Random thoughts about #nephrology #stats #bioinformatics #perl #HPC #nanopore #RNAseq #covid19 #amateurcoding
2,777 posts
9,436 followers
2,310 following
Regular Contributor
Active Commenter
comment in response to
post
It is the bad nephrologists' fault, the money driven scumbags who need incentive programs (and now bullets) to do the right thing /s
comment in response to
post
Εντωμεταξύ το κλαμπ στη μνήμη διαφήμιση έχει λογαριασμό στο τουητερ και είναι ορθόδοξο, Νασερικο, Ανδρεικο ΠΑΣΟΚ. Λέω να επισκεφτώ το καλοκαίρι
comment in response to
post
Inflammation, vascular reactivity, thrombosis are all interconnected and converge on good things acutely but are counterproductive when they are chronically activated
comment in response to
post
Να τους παίξεις τον βαθύτατα ερωτικό στον έλεγχο διαβατηρίων
@cubiclogic.bsky.social @madgreek.bsky.social
comment in response to
post
I gotta say, Grok is the only LLM i am impressed by simply because it so obviously defeats its maker & owner. I wonder how much of Musks reaction to it are on purpose to make people impressed by Groks independence
comment in response to
post
The singularity is here
comment in response to
post
comment in response to
post
Βάλτε Λίνουξ κ κόψτε τις αηδίες. Το μεγαλύτερο πρόγραμμα θα είναι τα win10 αν έχεις corporate πρόγραμμα που κατατάσσει υπολογιστές πάνω από κάποια ηλικία σαν πεπαλαιωμενους. Θα πρέπει να σκάσω 20χιλ να ξανα-αγορασω υπολογιστές για το τμήμα μου το Σεπτέμβριο
comment in response to
post
@stephensenn.bsky.social you may want to take a look at this. Fascinating statistical implications for selection into trials and analytical methods
comment in response to
post
The variable rate of progression means that trials will have to be huge to develop effects, and trad approaches to conducting clinical trials and analyzing their data will be an impediment. This field (similar to HIV) will require a qualified sensitive and specific biomarker to speed up developments
comment in response to
post
AI is powerful
comment in response to
post
Wait, are you chemtrail watching?
comment in response to
post
With non deterministic query results
comment in response to
post
I just challenge anyone to find interactions like this on linked in
comment in response to
post
Yes
comment in response to
post
😂😂😂😂😂
But I have the best titbot followers
comment in response to
post
Yes, that too!!!!
comment in response to
post
It was good, wasn't it?
comment in response to
post
🤣🤣🤣🤣🤣
comment in response to
post
They are neither a reliable machine, nor they have a mind
comment in response to
post
Nah this will kill you in analysis because of the joins. For transactional use cases, though between 4 to 5 is optimal
comment in response to
post
Stay with 3 for most tasks.
comment in response to
post
Yes
comment in response to
post
TY. I found Dr Winship website as well with the key papers as pdf (as it should)
comment in response to
post
From a practical perspective, I think going to town with mgcv (in #rstats) utilizing a tensor product basis over the APC space (or AC or PC or AP) while updating the participant age in the datasets and selecting clinically & policy relevant scenarios for lpmatrix predictions works!
comment in response to
post
1. Bounded effects should formally be put in any method (what's the ref to Fosse & Winship btw)
2. Allocating the drift requires subject matter knowledge, and I don't think a general approach would work IMHO
3. Practitioners should invest more time selecting appropriate references for predictions
comment in response to
post
Has anything really changed from Bendix Carstensen's nice paper from nearly 20 years ago?
pubmed.ncbi.nlm.nih.gov/17177166/
Constant and linear effects are not identifiable, but nonlinear terms are, and predicted rates are used to summarize stuff
comment in response to
post
Why stop at the rectum, when the big toe is clearly a candidate entry point to the meninges?