Sumant Ranji
sumantranji.bsky.social
Sumant Ranji
@sumantranji.bsky.social
160 followers 550 following 80 posts
Professor of Medicine at UCSF, hospitalist at San Francisco General Hospital. Fan of good music and bad teams. Views my own.
Posts Media Videos Starter Packs
Reposted by Sumant Ranji
When patients don't trust their health insurance, cancer diagnostic delays double (30%→60%). Coverage uncertainty itself disrupts care. Read more and watch the interview on this new research: codex.ucsf.edu/coverage-unc...

@ajsharmamd.bsky.social
Agree, but that’s the issue - “be thoughtful” means something different to everyone
Interestingly my residents last month were mostly anti-AI, in the sense that they use OpenEvidence to look things up but were very skeptical about any other uses. I told them that was the right attitude!
Reposted by Sumant Ranji
AI is transforming healthcare fast—but how do we ensure it actually improves diagnosis? UCSF CODEX brought together 30 healthcare leaders to develop practical metrics for measuring AI's impact on diagnostic excellence. Full story: codex.ucsf.edu/news/healthc...
I only partly understood your original explanation, so this question may be nonsensical. How then do reasoning models work differently from standard LLMs? Are they actually encoding new information as part of their "thinking" process?
Reposted by Sumant Ranji
Things today's White House autism announcement did not do:

-Discuss strategies to increase supports and services for autistic people/families
-Highlight why autistic people and families deserve respect, not pity or fear
-Endorse one speck of autism research that is not dangerous pseudoscience
Nope. The original just sounded like crap. Lo-fi has its charms but not for the ‘Mats and especially not for songs like these.
It is grimly ironic to write “I am concerned about MAHA” in the chart
Strongly disagree with you on this. The authors of the paper you cite include the Dean of the Harvard School of Public Health. They are hardly being gatekept. If anything RFK Jr and Trump’s crusade against Harvard is impeding their research far more than any alleged left wing suppression.
There's something worth investigating here although the paper's title is more definitive than feels warranted. My point is that RFK Jr and his well-funded, very skilled allies are using these studies to continue a campaign against autistic people. It is very dangerous to afford them any legitimacy.
He is very skilled at exploiting highly specialized and complex scientific disagreements as definitively "right" or "wrong" to further his own agenda - which fundamentally discriminates against women and autistic people
4/ 4th, untreated pain is very bad! All of medicine involves risks and benefits. Which is worse, a possibly very slightly increased risk of ASD or millions of pregnant people suffering through pregnancy even more?

This is what RFK Jr does and why he is so dangerous...
3/ 3rd, even if true, the magnitude of the association btw APAP and ASD/ADHD is much smaller than other known associations (genetics, prenatal stress, perinatal complications)
2/ Given how common APAP use is, teasing out the effect of APAP vs other known risk factors is very challenging.

2nd, there is no biologically plausible mechanism by which occasional APAP use causes ASD/ADHD. APAP is certainly very toxic at high levels or with chronic use, but a couple of pills?
1/ There are multiple issues here that cannot be separated from the fact that RFK Jr is a dangerous charlatan.

First, any association between acetaminophen and ASD/ADHD is likely confounding by indication. People take APAP to treat fever and chronic illnesses - which are themselves linked to ASD
Reposted by Sumant Ranji
I wrote a little thing about Chicago, my beautiful complicated city, on the eve of occupation. dansinker.com/posts/202…
thought it was "high wire in a fire representatives have engaged in a government for hire and a combat site"
Always thought it was “look at that low plane, fire, death”
Thank you! Can’t wait to hear the full show - a few songs have circulated for a long time. A Hüsker live album is long overdue
Huh, to me Stasium had a very light touch on LOTD - the drums sound better and Paul’s fade out vocals are more prominent but otherwise it sounds close to Chilton’s original production