Karandeep Singh
@kdpsingh.bsky.social
š¤ 4809
š„ 666
š 485
Faculty at UC San Diego.
#rstats
. Creator of Tidier.jl
#julialang
.
#GoBlue
. Views own.
pinned post!
No amount of AI can overcome bad health policy.
6 months ago
1
23
5
CDC: Disease is a human right.
about 3 hours ago
1
4
0
When in Rome, impute like the Romans.
2 days ago
0
4
1
āļø š±
2 days ago
2
6
0
Essential tools to do data science/prediction/simulation modeling in Python: * marimo notebooks (pip install marimo; marimo edit --sandbox ___.py) * ibis ("ibis-framework[duckdb]" to install it) * sklearn (with major emphasis on make_pipeline) * HSMA-style simpy (
des.hsma.co.uk
) + vidigi
loading . . .
HSMA - the little book of DES
An introduction to writing discrete event simulation models for healthcare (and beyond) in SimPy
https://des.hsma.co.uk
2 days ago
1
5
2
iOS 26 fails in many ways but not being able to instantly close a tab in Safari takes the cake for me as the worst part.
3 days ago
3
8
0
reposted by
Karandeep Singh
Tidier.jl
4 days ago
TidierData 0.17.2 is now available, bringing temporary grouping and improvements to `unnest_wider` available to a DataFrame near you.
github.com/TidierOrg/Ti...
0
4
1
My favorite paper company is __mifflin__
4 days ago
0
4
0
reposted by
Karandeep Singh
Mehmet Hakan Satman
7 days ago
SubsetJuliaVM: A Julia Subset Execution Environment for iOS and Web
#JuliaLang
discourse.julialang.org/t/subsetjuli...
loading . . .
SubsetJuliaVM: A Julia Subset Execution Environment for iOS and Web
Hello everyone. I want to introduce SubsetJuliaVM, a subset of Julia that works on iOS devices and web browsers. The following text was generated using https://translate.preferredai.jp/ to rewrite th...
https://discourse.julialang.org/t/subsetjuliavm-a-julia-subset-execution-environment-for-ios-and-web/134814?u=jbytecode
0
3
2
reposted by
Karandeep Singh
Elizabeth Jacobs, PhD
7 days ago
Iām seeing a ton of disinformation from the right wing about Californiaās new law requiring folic acid fortification for corn masa flour, so here are some facts (US only). 1. By law, wheat flour has been fortified with folic acid since 1998 to prevent neural tube defects like spina bifida.
15
341
130
reposted by
Karandeep Singh
Rohit Vashisht
9 days ago
Bhailang š A toy programming language written in typescript! One need to know college canteen level Hindi to use this.
bhailang.js.org
loading . . .
Bhailang - A toy programming language based on an inside joke
Bhailang is dynamically typed toy programming language, based on an inside joke, written in Typescript. Created by Aniket Singh and Rishabh Tripathi.
https://bhailang.js.org
0
5
2
Ordering coffee at a coffee place: Which rainforest would you like your beans ethically sourced from? Ordering chai at a coffee place: How much syrup? Ordering chai at a chai place: What ratio of cloves, cardamom, ginger, and fennel seeds do you like? Ordering coffee at a chai place: Hot or cold?
10 days ago
0
6
0
reposted by
Karandeep Singh
12 days ago
hello, world!
1
3
1
App: Do you like this app? User: Yes ā¦15 seconds pass⦠iOS: On a completely unrelated note, would you like to give this app a star rating?
12 days ago
1
5
0
Theyāre saying the quiet part out loud.
17 days ago
0
6
0
People worry about the AI bubble. AI worries about the people bubble.
17 days ago
0
2
0
reposted by
Karandeep Singh
NEJM AI
20 days ago
Editorial by Eileen Kim, MD, Vincent X. Liu, MD, MSc, and Karandeep Singh, MD, MMSc: AI Scribes Are Not Productivity Tools (Yet)
nejm.ai/48aLhvx
@kdpsingh.bsky.social
#AI
#MedSky
#MLSky
1
3
2
reposted by
Karandeep Singh
Brittany Partridge, MBA, FAMIA
20 days ago
The problem with virtual care was never the tools. It was the lack of shared rules for presence once proximity changed. Im calling that missing foundation the Presence Layer.
#Informatics
0
3
1
reposted by
Karandeep Singh
James Balamuta
21 days ago
Shiny for R: GPL-3 ā MIT ⨠Your dashboards are now legally chill. Deploy without consulting three lawyers and a ouija board first.
#rstats
#rshiny
3
20
3
"Crypto-Standard Science" explains a lot of what is happening right now. FDA requiring 1 clinical trial instead of 2 for drug approval? Crypto-Standard Science FDA proposing "impossible" standards for vaccine testing? CSS FDA holding closed-door meetings with Russell Brand? CSS
23 days ago
1
4
1
Personal view: āScientists no longer have to mouth DEI shibboleths to garner fundingā is an interesting way of saying that the NIH does not plan to address health disparities and the health of minorities.
thespectator.com/topic/cured-...
loading . . .
How we cured DEI at the National Institutes of Health
The National Institutes of Health will no longer fund research based on DEI that has diverted resources from advancing health
https://thespectator.com/topic/cured-dei-national-institutes-health-jay-bhattacharya-matthew-memoli/
24 days ago
1
15
1
reposted by
Karandeep Singh
Frank Harrell
25 days ago
Nice paper Ben but it's hard to justify use of AUROC except when there is only one candidate model. AUROC should not be used to judge improvements in predictive discrimination. The width of a well-calibrated risk distribution is far better for that.
#StatsSky
#Statistics
fharrell.com/post/addvalue
loading . . .
Statistically Efficient Ways to Quantify Added Predictive Value of New Measurements ā Statistical Thinking
Researchers have used contorted, inefficient, and arbitrary analyses to demonstrated added value in biomarkers, genes, and new lab measurements. Traditional statistical measures have always been up to...
https://fharrell.com/post/addvalue
1
4
3
reposted by
Karandeep Singh
Ben Van Calster
25 days ago
Our guidance regarding performance measures for medical AI models is finally out! - Stop bashing AUROC, although it does not settle things - Calibration and clinical utility are key - Show risk distributions - Classification statistics (e.g. F1) are improper
www.thelancet.com/journals/lan...
loading . . .
Evaluation of performance measures in predictive artificial intelligence models to support medical decisions: overview and guidance
Numerous measures have been proposed to illustrate the performance of predictive artificial intelligence (AI) models. Selecting appropriate performance measures is essential for predictive AI models i...
https://www.thelancet.com/journals/landig/article/PIIS2589-7500(25)00098-6/fulltext
2
48
25
reposted by
Karandeep Singh
Sung Kim
25 days ago
Luxical: A new text embedding model that excels in CPU. Blog:
datologyai.com/blog/introdu...
Paper:
arxiv.org/abs/2512.09015
Code:
github.com/datologyai/l...
Model:
huggingface.co/DatologyAI/l...
1
27
4
reposted by
Karandeep Singh
Emily Riederer
25 days ago
Officially stashing away āTheir trajectory is parallel to mine on a log scale at leastā for any time comparison in robbing me of job in life (This is a post on insecurity not a knock on rstats if that wasnāt clear)
1
11
1
reposted by
Karandeep Singh
NEJM AI
26 days ago
The authors of a new editorial write that
#AI
scribes for clinical practice show early promise for improved productivity and lay out the conditions under which these tools may substantially improve productivity in the future. Read the full editorial:
nejm.ai/48aLhvx
@kdpsingh.bsky.social
3
2
3
AI Scribe Called Quest is the crossover album we all deserve but will never see.
26 days ago
0
5
1
reposted by
Karandeep Singh
Kelly Bodwin
29 days ago
%notin% is coming to Base R! Heck to the yes. We are truly blessed on this day, thank you R Core. š¤©
#rstats
4
55
9
reposted by
Karandeep Singh
Jacobs Center for Health Innovation @UC San Diego Health
29 days ago
Catch JCHI Exec Director
@kdpsingh.bsky.social
@ucsdhealth.bsky.social
Dec 10, 1 p.m. PST for a virtual keynote at Stanford AI+HEALTH Conference. A great lineup on the latest in AI and healthcare. Register ā¬ļø
medhub.stanford.edu/event/aiheal...
0
5
2
The Lesser-Known StreisAInd Effect Statistician: AI gets too much hype. Iām going to fix that. ML researcher: How? Stats: Iām going to give a talk about AI hype⦠ML: ⦠Stats: and then do a press release about it⦠ML: To hype it up? Stats: Yes.
29 days ago
0
3
0
reposted by
Karandeep Singh
Andrew Vickers
30 days ago
Machine learning has developed remarkable new ideas about how to develop prediction algorithms. But always baffled me why the field had to reinvent how to evaluate models. Here we show F score should not be used to evaluate medical prediction models
link.springer.com/epdf/10.1186...
loading . . .
The F score ranks diagnostic tests and prediction models inconsistently with their clinical utility
https://link.springer.com/epdf/10.1186/s41512-025-00214-7?sharing_token=GoiCZejO_nrt-7oypPf6Zm_BpE1tBhCbnbw3BuzI2ROUQWl3TFGd0ArK1wLS-Jm0Jh5y5UAQmXhhB3R1t1UzskLHVAk2QLw3kzFEg8S6miBLhX6awlMLya_uBKno_bPaFGMXHD_tV6eLtlm7w8ctITrNO2Lf6wdN9L-4jj5pwrg%3D
1
12
5
iOS 26: āYou are now entering Game Mode. Iām going to just leave this banner up here for 20 seconds to make sure you are aware. Apologies if this completely blocks a critical part of the screen as you play.ā iOS 26: āWait, why did you just quit the game?ā
about 1 month ago
0
2
0
reposted by
Karandeep Singh
Vijay G. Sankaran
about 1 month ago
Really enjoyed the fantastic session we had yesterday on applications of AI in hematology at
#ASH25
with
@wondermixtape.bsky.social
, Sanam Loghavi, and
@kdpsingh.bsky.social
!
0
13
2
reposted by
Karandeep Singh
Shannon McWeeney
about 1 month ago
Looking forward to our joint session Advancing Hematology Through Artificial Intelligence today at 4pm Hyatt Regency Ballroom S
#ASH25
w/ Vijay Sankaran Sanam Loghavi &
@kdpsingh.bsky.social
on how
#AI
is will impact hematology across discovery science, diagnostics, & clinical care
0
2
1
reposted by
Karandeep Singh
Tim Kellogg
about 1 month ago
TITANS & MIRAS: real continual learning MIRAS = a unifying theory of transformers (attention) and state space models (SSM, e.g. Mamba, RNNs) TITANS = an optimal MIRAS implementation thatās āhalfway betweenā SSM & transformer with a CL memory module letās dive in!
research.google/blog/titans-...
loading . . .
Titans + MIRAS: Helping AI have long-term memory
https://research.google/blog/titans-miras-helping-ai-have-long-term-memory/?utm_source=twitter&utm_medium=social&utm_campaign=social_post&utm_content=gr-acct
1
42
12
reposted by
Karandeep Singh
Ben Harrap
about 1 month ago
%notin% coming soon to an
#rstats
environment near you š¤©
developer.r-project.org/blosxom.cgi/...
loading . . .
Daily News: R-devel/NEWS
https://developer.r-project.org/blosxom.cgi/R-devel/NEWS/2025/12/05
1
103
25
reposted by
Karandeep Singh
Andrea
about 1 month ago
Round of applause for this researcherās work. Brilliant backdoor attack. 𤣠Kohler Health also confirmed that it can harness the collected poop š© pics to train AI.
1
1
1
arstechnica.com/health/2025/...
loading . . .
12 former FDA chiefs unite to say agency memo on vaccines is deeply stupid
Prasadās arguments āmisrepresent both the science and the regulatory record.āā¦
https://arstechnica.com/health/2025/12/12-former-fda-chiefs-unite-to-say-agency-memo-on-vaccines-is-deeply-stupid/
about 1 month ago
0
2
0
āGold-standard scienceā
arstechnica.com/health/2025/...
about 1 month ago
0
1
0
This gives āend-to-endā encryption a whole new meaning.
arstechnica.com/gadgets/2025...
loading . . .
Engineer proves that Kohlerās smart toilet cameras arenāt very private
Kohler is getting the scoop on peopleās poop.
https://arstechnica.com/gadgets/2025/12/despite-accessing-user-data-kohler-still-says-its-smart-toilet-cameras-use-e2ee/
about 1 month ago
1
6
0
reposted by
Karandeep Singh
Scott McGrath
about 1 month ago
@kdpsingh.bsky.social
had an editorial that also discusses these findings. Important studies on Ambient AI for clinicians.
#MedSky
add a skeleton here at some point
1
3
3
reposted by
Karandeep Singh
Brittany Trang
about 1 month ago
We finally have randomized, controlled clinical trial results from ambient AI scribes. But what do these RCTs actually tell us? Look at the below chart. Time savings are often low, but doctors give the technology rave reviews, esp in anecdotes that people tend to believe more than data š š§µ š„ļøš©ŗ
1
25
23
reposted by
Karandeep Singh
Nick Byrd, Ph.D.
about 1 month ago
Can doctors save time and stress with
#AI
scribes who listen to interactions with patients and create medical notes? After many non-randomized studies without control groups, there are now results from two randomized control trials:
1
14
4
reposted by
Karandeep Singh
Rohit Vashisht
about 1 month ago
Excellent study by team
@kdpsingh.bsky.social
. Using statistical models, streamlined versions of surgical supply lists can reduce hospital waste and improve operational efficiency. An inspiring use of data to reduce waste without compromising care! Way to go!
health.ucsd.edu/news/press-r...
loading . . .
Data-Driven Surgical Supply Lists Can Reduce Hospital Cost and Waste
UC San Diego study shows data-driven surgical supply lists can cut waste, save millions and improve efficiency without compromising patient safety.
https://health.ucsd.edu/news/press-releases/2025-11-26-data-driven-surgical-supply-lists-can-reduce-hospital-cost-and-waste/
1
3
2
reposted by
Karandeep Singh
Malcolm ęē²¾ Barrett
about 1 month ago
Linear in the log scale anyway. You backtransform them and the whole place falls apart
1
5
1
Most log cabins are actually linear cabins.
about 1 month ago
1
15
0
reposted by
Karandeep Singh
Tyler Morgan-Wall
about 2 months ago
Was going through my video archives looking for something and laughed at this, thought I'd let Bluesky enjoy
#RStats
#rayshader
loading . . .
1
26
5
"Do AI scribes actually help?" Figure 1:
about 2 months ago
1
5
0
reposted by
Karandeep Singh
Elizabeth Jacobs, PhD
about 2 months ago
This is completely disgusting. Today, the CDC updated their vaccine safety page. It now says: āThe claim "vaccines do not cause autism" is not an evidence-based claim because studies have not ruled out the possibility that infant vaccines cause autismā.
147
2159
780
Alex, Iāll take Stories Not About AI But About AI for 1000.
arstechnica.com/science/2025...
loading . . .
How Louvre thieves exploited human psychology to avoid suspicionāand what it reveals about AI
For humans and AI, when something fits the category of āordinary,ā it slips from notice.
https://arstechnica.com/science/2025/11/how-louvre-thieves-exploited-human-psychology-to-avoid-suspicion-and-what-it-reveals-about-ai/
about 2 months ago
0
6
0
reposted by
Karandeep Singh
Ryan Radecki MD MS
about 2 months ago
The perpetual interface between academia and the real world.
add a skeleton here at some point
0
4
1
Load more
feeds!
log in