The brain can be modeled as a generative system that predicts bodily and environmental input and reduces mismatch through updating and action. From this perspective, stress is treated not as prediction error per se, but as a condition in which mismatch persists without sufficient recalibration. Health is therefore defined not as the absence of symptoms, but as the preservation of recalibrability, whereas psychopathology is conceptualized as the progressive fixation of recalibration failure. This paper proposes ESMR as a four-layer, two-timescale process framework for comparing psychosomatic conditions, psychiatric disorders, and psychosis-level phenomena in terms of partially shared failures of recalibration rather than as a single disease continuum. ESMR distinguishes E (Embodied Constraints), S (Salience Calibration), M (Metacognitive Model Revision), and R (Reality-Grounding Interface) across learning (L) and development (D). Development is treated not merely as slower accumulation of prior learning, but as change in the biological, representational, and self-regulatory architecture that alters both what forms of recalibration are possible and the background conditions under which learning can be taken up. A central distinction is that M concerns whether corrective information can be converted into updating, whereas R concerns whether already understood correction can be adopted as self-relevant reality under embodied, practical, and interpersonal constraints, with downstream transfer treated as a consequence rather than part of the R core. The framework predicts layer-specific cross-signatures rather than a single undifferentiated burden effect and outlines a staged empirical program using framing, disconfirmation, reality-grounding, and embodied/interoceptive tasks. ESMR is offered as a revisable, clinically oriented intermediate framework rather than a completed computational model.
AI-guided redesign of laboratory-evolved reverse transcriptases enhances prime editing
Nature Biotechnology, Published online: 21 May 2026; doi:10.1038/s41587-026-03149-6
A computational redesign strategy improves evolved prime editors.
Integrating public health into house design
Nature Medicine, Published online: 21 May 2026; doi:10.1038/s41591-026-04413-7
Housing is not just a backdrop to health, but a powerful intervention. A study now shows that redesigning the home environment can improve child health outcomes, but success requires community co-design and solutions for cost and scalability.
Resetting autoimmune disease with CAR cell therapies
Nature Medicine, Published online: 21 May 2026; doi:10.1038/s41591-026-04430-6
This Review surveys the fast-developing landscape of CAR T cells in autoimmune disease from both a technological and clinical standpoint, exploring the concept of deep B cell depletion and immune ‘reset’.
Foamy microglia link oxylipins to disease progression in multiple sclerosis
Nature Neuroscience, Published online: 21 May 2026; doi:10.1038/s41593-026-02302-3
Foamy microglia are associated with multiple sclerosis progression, linking phagocytosis, altered lipid metabolism (oxylipins) and lysosomal stress to nonclassical neuroinflammation and disease severity, highlighting potential therapeutic targets.
STAT+: RFK Jr.’s screen time warning
You’re reading the web edition of D.C. Diagnosis, STAT’s twice-weekly newsletter about the politics and policy of health and medicine. Sign up here to receive it in your inbox on Tuesdays and Thursdays.
The Pitt star Noah Wyle is at the Capitol this week pushing for legislation to give tax credits to health care providers in staff-shortage areas, among other health care bills. Send news tips and your favorite Pitt episodes to John.Wilkerson@statnews.com or John_Wilkerson.07 on Signal.
MAHA is mad about this Trump win
When Sen. Bill Cassidy (R-La.) lost his primary, MAHA rejoiced. Not so with Rep. Thomas Massie, the Kentucky Republican who lost against a Trump-backed challenger.
STAT+: Immunovant shares surge on arthritis trial data
Want to stay on top of the science and politics driving biotech today? Sign up to get our biotech newsletter in your inbox.
Gene therapy delivered before birth may finally be inching toward reality, as UCSF researchers have submitted an FDA application to target a devastating neurodegenerative disorder that kills many children before age 3.
Also, the FDA leadership tumult is rattling investors, destabilizing rare disease drug development, and raising concern about the agency’s scientific independence.
STAT+: Acting head of NIH’s infectious disease institute reported to have stepped down
Yet another leadership position at the National Institutes of Health appears to be vacant. Jeffery Taubenberger, who has been serving as acting head of the National Institute for Allergy and Infectious Diseases, has stepped down, Sen. Tammy Baldwin revealed Thursday during a hearing of the Senate Appropriations Committee.
It was unclear when Taubenberger, who has been serving in the role since April 2025, stepped down, or why, though there has been chatter in infectious diseases research circles that Taubenberger had stepped down about two weeks ago. STAT has asked the Department of Health and Human Services about his status several times; those queries have gone unanswered and unacknowledged.
Taubenberger was still listed as acting director on the institute’s website on Thursday. In the HHS employee directory he is listed as the chief of the viral pathogenesis and evolution section. He has not responded to repeated queries from STAT about his status.
STAT+: Elevance executive ordered to testify in Medicare Advantage fraud case
Peter Haytaian, a former top executive at health insurance company Elevance Health, will have to sit for a deposition in the federal government’s case that alleges Elevance committed fraud in its Medicare Advantage plans, a judge ruled late Wednesday.
The decision is a victory for the Department of Justice, which initiated the highly watched case six years ago and has sought to interview Haytaian since February. It also creates a rare chance for a high-ranking official to discuss, under oath, how a Medicare plan operates.
Elevance has blocked the DOJ from accessing Haytaian for a deposition, arguing he “lacks personal knowledge of the business practices at issue” even though he oversaw Elevance’s government health plans for more than four years — which included the period when Elevance allegedly overbilled the government for its Medicare Advantage plans.
Anthropic’s Code with Claude showed off coding’s future—whether you like it or not
The vibes were strong at Code with Claude, Anthropic’s two-day event for software developers in London that kicked off on May 19, the same day as Google’s I/O in Palo Alto. (A coincidence, not a flex, Anthropic staffers assured me.)
“Who here has shipped a pull request in the last week that was completely written by Claude?” Jeremy Hadfield, an engineer at Anthropic, asked from the main stage. Almost half the people in the packed room—many sitting with laptops on their knees, coding or prompting as they watched the talks—raised their hands.
Pull requests are fixes or updates to existing software that are submitted for review before they go live. They are the bread and butter of software development, the chunks of code that most professional developers spend their lives writing—or did until now.
“Who here has shipped a pull request that was completely written by Claude where they did not read the code at all?” Hadfield asked next. Nervous laughter. Most of the hands stayed up.
It’s not news that LLM-powered tools like Anthropic’s Claude Code and OpenAI’s Codex have upended the way software gets made. Top tech companies now like to boast of how little code their developers write by hand. (“Most software at Anthropic is now written by Claude,” Hadfield said. “Claude has written most of the code in Claude Code.”) OpenAI, Google, and Microsoft make similar claims. Many others wish they could.
Even so, it is striking how normal this new paradigm already seems, and how fast it has set in. This was the second year that Anthropic has put on developer events, which also run in San Francisco and Tokyo. This time last year, the company had just released Claude 4. It could code, kind of. But with Anthropic’s latest string of updates—especially Claude 4.6 and then 4.7, released in February and April—Claude Code is a tool that more and more developers seem happy to hand their work off to.
ANTHROPIC (GRAPHIC) / WILL DOUGLAS HEAVEN (PHOTO)
Anthropic says its goal is to push automation as far as it will go. Instead of using AI to generate code and then having humans clean it up and fix the mistakes, it wants Claude to check and correct its own work. “The default isn’t ‘I’m going to prompt Claude’—the default is now ‘I’m going to have Claude prompt itself,’” Boris Cherny, who heads Claude Code, said in the opening keynote.
If all goes well, human developers shouldn’t even see the error messages when something doesn’t work. That will all be handled by Claude, which will test and tweak, test and tweak, until everything runs as it should. As Ravi Trivedi, an engineer at Anthropic, put it in another talk: “The key principle is getting out of Claude’s way. We like to say: ‘Let it cook.’”
Trivedi presented a new feature in Claude Code, announced two weeks ago, which Anthropic calls dreaming. Claude Code agents write notes to themselves, recording and saving useful information about specific tasks. When another coding agent later starts to work on the same code, it can use the notes to get up to speed faster and learn from any errors that previous agents may have made.
Dreaming is a system that Claude Code uses to read through all these notes and consolidate the information they contain, spotting patterns and common issues across different tasks. In theory, dreaming should help Claude Code learn about a particular code base and get better and better at working on it.
Success stories
Code with Claude is an event aimed at developers. As well as product showcases and hands-on workshops from Anthropic, there were how-tos from a range of companies that had reshaped their software development teams around Claude Code, including Spotify and Delivery Hero as well as Lovable, Base44, and Monday.com—three startups vibe-coding apps that help people vibe-code apps.
There were no signs of unease at Code with Claude. Everybody I met wanted in.
And yet outside the conference there have been a number of reports that many coders are starting to question this bright new future. Some gripe in online forums like Reddit and Hacker News that AI coding tools are being pushed by managers chasing productivity gains, when in practice the technology makes software development harder because of all the extra code developers now have to review. “The only people I’ve heard saying that generated code is fine are those who don’t read it,” a user called pron posted on Hacker News last week.
Others claim that their coding abilities have fallen off as they hand more tasks to AI. And researchers have warned that AI tools can produce unsafe code that will make software more vulnerable to attacks.
I sat down with Claude engineering lead Katelyn Lesse and Claude product lead Angela Jiang and asked them what they made of the concerns that a sudden flood of code generated (and shipped) without proper human oversight was kicking serious security and maintenance problems down the road.
“All of the old software development best practices still apply. They’ve applied this entire time,” said Lesse. “I think there are a lot of people and teams that may have lost sight of them in this moment.”
And yet as Anthropic and others push for greater automation and tools like Claude Code improve, the temptation increases to offload more and more tasks, including oversight. Lesse told me that some of the technical managers at Anthropic are exhausted by keeping up with all the code their teams now produce. “Part of things happening so much more quickly is just managing your time,” she said.
“I think that right now Claude is probably as good as a midlevel engineer at writing code,” she added. You still need expert engineers to design a system and troubleshoot harder problems, she said, “But over time we want Claude to get better and better at all different types of engineering.”
Jiang agreed: “I think the absolute end state we’re trying to get to is Claude basically being able to build itself.”