The US Administration has embarked on a large, unprecedented program of excluding lawful, highly-skilled workers and students from the United States.
I talked with the brilliant @gracesegers.bsky.social at @newrepublic.com about the broad economic ripple effects.
newrepublic.com/article/2073...
Despite recent controversial decisions, the FDA has stopped holding advisory committee meetings — which bring together regulators, companies, patients, and a panel of independent advisors to publicly discuss complex regulatory matters. I dug into why:
www.statnews.com/2026/03/09/f...
🎉March/April Pulse Highlight🎉
Where @lsgeriatricem.bsky.social talks about her Fulbright experience in Australia!
issuu.com/saemonline/d...
Pages 66-67
@saemonline.bsky.social
#emergencymedicine #saem #agem #geriatricmedicine #emergencydepartment
The data also reinforce a message that should be clear: American leadership in science and medical research requires far more than getting across the line on top-line budget negotiations and will require continued vigilance to track how HHS treats applications and existing awards.
The plot is only part of the story.
The data do not include lost science via terminations, freezes, stalled payments, and award delays. The collapse in awards rates will lead to job loss, gaps in research programs, and drive scientists to spend more time writing grants rather than doing science.
Graph of award probability of R35 and R01 from NIH factbook as a function of review rank percentile. As is apparent, 2025 is a significant departure, with lower award probabilities at all scores <40 and significant departures from norm, where even being in the top 10% is no longer a nearly certain indicator of success. Data source: https://report.nih.gov/nihdatabook/report/302
The data is in: the NIH goalposts have shifted.
What were once almost certain fundable scores have become coin flips and what used to be likely grants have become aspirational, leading to fewer awards.
Another manifestation of how HHS policies have led to fewer awards and less science.
This is not just cursed, its monstrous. The digital resurrection of a historian who died in January of this year, all so Grammarly can get some more clicks and engagement from students and/or scholars and/or others.
It feels so wrong on so many levels, these ghosts enslaved to AI forever
Luigi Guido speaking on The Economic Costs of Ambiguous Laws
Would love to learn more here if you have references to what Claude is doing around fact verification.
LLMs reporting on their own reasoning are generally not particularly reliable. Would absolutely be possible to incorporate other models on top of an LLM, although the state of the science there is in flux. But at that point, it's not an LLM, it's some kind of composite model?
"Structurally indifferent to truth is a good line" is a good line. Also worth noting that not all models are, just that LLMs weren't even designed to have a concept of truth. One of many ways the "AI" moniker distracts.
The critique of unmeasured confounding is often levied in a lazy/broad way. It is trivially true in any observational study. But if the critic can't think of a plausible such confounder and posit a reasonable direction/magnitude of its bias then they're not doing productive science.
Anakin-Padme meme. Anakin: Additionally, we considered the confounders c1, c2 and c3. Padme: And you included all of them in a single model, right? Padme: And you included all of them in a single model, right?
Psychologists following a third-variable control strategy that may be best described as "sequential."
Join NYT's @jeneeninterlandi.bsky.social, Dr. Jonathan Mannheim, and me on Monday, March 2nd at 1pm ET, Live on Substack as we discuss the human cost of the administration's war on US science.
Link to Live: open.substack.com/live-stream/...
Info below:
insidemedicine.substack.com/p/speaking-o...
A line graph showing NSF grant awards made through 2/27/26 for fiscal year 2026 compared with grant awards for fiscal years 2021-2025.
NSF Update (Awards through 2/27/26)
Directorates to follow
1/10
The impact is already hitting science agencies hard. The NIH—the world's largest public biomedical research funder—has had to rely on leftover stopgap funds.
New grant awards have slowed to a trickle — exacerbating the effects of a record-long shutdown in October.
(h/t @jeremymberg.bsky.social)
That's right. Worse than Pointless
www.nytimes.com/2026/02/23/o...
It’s possible to hope the ayatollahs regime will fall one day and Iranians be free and be dismayed by our country bombing Iran with no probable cause , with no public debate , and no congressional authorization.
There were HUGE differences between patients who did and did not receive the AI stethoscope. Patients were 20 years older (64 vs 44), 3.5x more likely to have hypertension (53% vs 15%) and 4x more likely to have kidney disease (13% vs 3%)
🧵 11/
But per-protocol analyses are difficult. Whether a doctor chooses to use the "AI stethoscope" is not determined at random.
This 'breaks' the randomisation, making the design equivalent to an observational study.
🧵 6/
First, the headline should be the result of the primary analysis (the intention to treat), which reports "Implementation of an AI stethoscope in routine primary care did not significantly increase detection of heart failure or increase community-based diagnosis"
🧵 2/
Part of the problem here is a long-standing issue of mine: The public gets a huge benefit from beat journalism—but the journalism industry does not, by and large, recognize beat journalism for its value or when it comes to awards and prizes. Those go to investigations and "one big thing" features.
Authoritarians launch coercion against the Other, then expand it to political opponents and people with resources they want to control.
White Americans are used to govt respecting our rights. Good, Pretti, & 1000s of other Minnesotans learned this administration does not.
bsky.app/profile/jaya...
The administration is trying to aggressively escalate its coercive power over the American people.
Undiagnosed cognitive impairment is common in older patients discharged back to home. I’m interested in seeing the follow up results on how these patients manage recovery from the injury or illness that brought them to the ED. Great work team!
In case you were wondering how good we are at detecting cognitive impairment in the ED absent a systematic approach to do so, here are clock draws from 63 patients discharged from a single ED.
In @annalsofem.bsky.social :
www.annemergmed.com/article/S019...
Yesterday, those who teach Intro to Sociology at Florida colleges (as opposed to universities) received a ready-made curriculum from the state and were ordered to teach it.
Yes, you read that correctly. The *state* is enforcing a curriculum on college profs, complete w/ the following restrictions:
a graph showing how LLMs do or do not p-hack
While LLMs will try to follow good research practices by default, you can pretty easily convince them to p-hack for you. In one case (out of the 4 tested), the LLM moved the result from p > 0.05 to p < 0.001. github.com/janetmalzahn...
The high costs of isolationism
This is a vivid, horribly relevant read. “She does not sleep well. And as you lift up her pajama top to check her rash one morning, you see that her breathing is labored, shadows pooling between her ribs when she sucks in air. You suffer an icy moment of realization: This is a medical crisis.”
