Week of July 6, 2026
When I look at the arc of my July Fourth parade reflections (across now 6 years of newsletters), I see a slow conversion. I began as a skeptic. "I do not fully understand the appeal of parades," I wrote in 2021. By last summer, I had softened into grumpily supporting my family's enjoyment of the day through, in part, a sincere appreciation for the motorized porta-potties that appeared in our 2025 parade (less menacing than robotic dogs). This year tested that loyalty. It was a sweaty pageant of the nation's most earnest and least flattering instincts.
The heat made a persuasive case for scheduling your nation's Independence Day on a nice spring or crisp fall weekend (next time, perhaps we can time our revolution for better weather?). And yet, what I attended was less of a parade and more of an intellectual buffet of ideas, agendas, and advocacy. Landscapers, contractors, the Elks, the Moose, volunteer fire companies, the local Falun Gong contingent, politicians, the Ravens' marching band, a squad of alumni cheerleaders from the long-departed Baltimore Colts (a surprisingly spry group of septuagenarians), and an honor guard of older, kilted men with obviously painful hips leading bagpipers under Scottish flags. Aligned with our national Zeitgeist, I overheard security scolding a County Executive candidate for throwing candy at eager children (someone had to address the epidemic of hard-candy head injuries), while the Boumi Shriners, who raise money for sick children, drove mini trucks towing cages of stuffed animals.
I keep showing up to these parades looking for Norman Rockwell and a decent precision lawn-chair team. Every year, America hands me something stranger and more itself. I am still just trying to understand it.
See the Boumi Shriners pulling caged stuffed animals (what does this symbolism mean?!?): https://share.icloud.com/photos/0616Pq2Wdqw1o9b0p06KbHTjQ
I am traveling next weekend. The next newsletter will be on 7/20/26
The Google Notebook LM AI-generated podcast version of this week’s newsletter.
Science and Technology Trends
Scientists at the University of Minnesota built a synthetic, cell-like system entirely from purified, non-living biological components. The system includes a 90,000-base-pair genome spread across seven DNA plasmids within a lipid membrane. Across multiple generations, the “cell” could grow, copy its genome, divide, and allow a faster-growing variant to outcompete a slower one. Every previous attempt at a cell-like system started from the top down, stripping a living microbe down to the smallest genome that still functions. This is the first time a cell has been built up from non-living parts and run through a full cycle.
A few caveats: This is a preprint paper, not yet peer reviewed. No outside lab has reproduced it. The “cell” required hand-fed enzymes, E. coli ribosomes, and a protein trigger to divide. After five generations, only about 30 percent of “cells” still carried the complete genome. The senior author co-founded Biotic, the nonprofit publishing this work, and has filed a provisional patent. (Biotic looks like a genuine open-source effort rather than a for-profit organization.) All told, this could be a real milestone, but it needs confirmation in other labs. On the other hand, it may turn out to be more hype than the first step into functional, scalable biotechnology.
Interestingly, the primary researcher called these droplets "SpudCell" because she is Polish and thought they looked like potatoes. So, her team built a novel, potato-looking pseudo-organism that grows, divides, and dies on schedule, but only when fed “lunch.”
Consumer-level article
Paper
Spud Cell discussion from the biomedical public-benefit organization run by the researchers who published this work. It appears to be an organization to “open source” bioengineering advances:
AI-Assisted review of the pre-release paper
Last week, NIH released All of Us, a US-based biologic database (similar to the UK Biobank) containing data from more than 747,000 individual participants who (so far) have contributed 535,000 whole genomes linked to 482,000 electronic health records, along with narrative medical notes, diagnoses, and test results. The database also bundles the genetic information with health surveys about socioeconomic factors and location-based exposure data, such as air quality. All of Us’s funding comes from the 21st Century Cures Act, which is set to expire at the end of fiscal year 2026.
Alicia Martin, a statistical geneticist at the Broad Institute who already uses data from All of Us to build and test risk prediction tools, said: “[these data] offer opportunities to understand not just who is at risk of disease, but also who is more likely to progress or have some exacerbated health condition, and who is going to respond to specialized treatments.”
Sadly, the program’s budget has already been reduced by 72 percent since 2023. NIH released it just in time to see an enormously valuable data set fall victim to budgetary reductions in scientific research.
NY Times Article:
Program Website:
Since 2019–2020, the sodium-glucose transporter 2 (SGLT2) inhibitor class of medications (marketed as Farxiga, Invokana, and Jardiance) has become an important part of the treatment of patients with heart failure and chronic kidney disease. These medications were originally introduced to treat diabetes in the early 2010s. They offer only a modest improvement in blood sugar control, but between their introduction and 2020, clinical outcomes data overwhelmingly showed that SGLT2 inhibitors provide substantial protection against heart failure (fluid overload) and progressive loss of kidney function. The mechanism of kidney protection is reasonably well established. The mechanism of the heart failure protection is not.
Last week, a new analysis of one proposed explanation for that heart-failure benefit turned up across my medical-oriented social media feeds: a meta-analysis examining how SGLT2 inhibitors reduce the fat around the heart, called epicardial adipose tissue, or EAT.
This topic is not so novel, but it highlights a good example of how science actually evolves. We introduce a new drug for one purpose; it turns out to improve outcomes across a far wider range of conditions, and then we have to go back and figure out why. In the meantime, these medications have become mainstays for treating both kidney disease and heart failure, even though we don't fully understand how they work. It is not unusual in healthcare to see an observed benefit, only to have the mechanism reverse-engineered later.
If reducing epicardial fat is really how these drugs help the heart, then other molecules that shrink that fat pad may also generate the benefit. Keep in mind that EAT reduction is one of several competing explanations for SGLT2’s heart-failure benefit, but the one with the weakest link to clinical outcomes.
Article (from a less prestigious subset of the Nature journals)
AI-assisted review of the meta-analysis:
Anti-Anti-Science
It’s anti-science tapas this week, intellectual small plates that improve when accompanied by a fortified wine:
A large, two-decade study in Hong Kong is the latest to find NO link between use of acetaminophen (Tylenol) during pregnancy and autism or attention-deficit/hyperactivity disorder. No matter the trimester the drug was prescribed, the dose, or the frequency of use, no association between the medication and the later diagnosis in a cohort of 708,020 mother-child pairs in Hong Kong (with approximately 43.3% with prenatal paracetamol exposure).
Consumer-level article:
JAMA Article:
Science communicator who writes about this (amongst other topics):
From the Washington Post: FDA regulators, posting notices of upcoming meetings, stated that there isn’t enough evidence to allow certain peptides to be produced by compounding pharmacies, despite recent public comments by HHS Secretary Robert Kennedy. The upcoming meeting of the FDA’s Pharmacy Compounding Advisory Committee will consider:
- BPC‑157 for ulcerative colitis.
- KPV for wound healing and inflammatory conditions.
- TB‑500 for wound healing.
- MOTS‑c for obesity and osteoporosis.
The Post notes that at least seven of the members have ties to peptide-related businesses and clinics. Another member is the son of a congresswoman who has urged Kennedy to convene the panel. It is not unusual to have technical experts and industry representatives on these kinds of committees; however, it is hard to find a precedent for this degree of conflict of interest in membership. Though the committee is advisory, I am sure Polymarket odds on the outcome of this panel will favor approval of these compounds.
Washington Post Article
BMJ on the Conflict of Interest
FDA public meeting notice
More from my reading on the politics of sunscreen. Brian Underwood, the beauty and grooming director for Men's Health and Women's Health, examined the anti-sunscreen movement, its ideas, and the shaky foundations underneath them.
What he gets right:
- He draws the correct distinction between the movement's "endocrine disruption" alarm, which rests on the enormous blood levels forced into rodents and fish in high-dose studies, and the far lower levels actually absorbed in ordinary human use.
-
He interviewed dermatologists and scientists (who formulate these products), who offered more balanced than either camp's talking points.
What he leaves out:
-
The series does not address the FDA's June 2026 approval of bemotrizinol, an organic UV filtering-compound used in European and Asian sunscreens for two decades. The agency added it to the over-the-counter monograph (i.e. approved) at concentrations up to 6% and handed DSM, the manufacturer, 18 months of marketing exclusivity. However, this is the first new sunscreen compound the FDA has approved since the late 1990s, which says more about the pace of American sunscreen regulation than about the molecule and is reflective of this administration's push toward faster innovation and approvals.
My ongoing learnings from sunscreen fear mongering on social media:
- Sunscreen works. It reduces UV-induced skin damage and consistently lowers the rates of squamous cell carcinoma and pre-cancers, with more limited evidence for melanoma. The long-term questions about chronic absorption remain open, but there is still no signal of toxicity.
-
Many of these compounds are absorbed into the body, and what that absorption does is unclear. Claims of perfect safety and claims of mass endocrine disruption are both inflated. The data does not support either direction. However, and like many aspects of healthcare, we are going on the best data we have available. If you are fair-skinned, likely to get burned, and spend time outside, sunscreen offers a good way to protect yourself from a known outcome (repeated burns and skin cancer). I think it is foolish to avoid sunscreen to protect yourself from an unquantifiable theoretical risk (such as endocrine disruption or other toxicities).
-
Whatever the process got bemotrizinol approved, this is a win for consumers. It adds another UV-blocking agent, and it is among the least systemically absorbed organic UV compounds we have.
Men's Health article:
https://apple.news/Al_wjzQb4ROWZPjtApoUj4g
AI summary:
AI summary:
https://claude.ai/public/artifacts/5e63ce94-324f-47b1-ae0a-a57681e20e4b
A review of the medical literature on the organic UV-blocking compounds, including knowns and unknowns: https://www.openevidence.com/ask/34a11c61-e868-43e9-aa7a-e5a4287c3b1b
A dermatologist's take (Matt Zirwas), well written and reasonably sensible: some sun is probably good, too much avoidance may be bad, too much sun is also bad, and if you are outside for long stretches, wear sunscreen:
A review of the medical literature on the organic UV-blocking compounds, including knowns and unknowns: https://www.openevidence.com/ask/34a11c61-e868-43e9-aa7a-e5a4287c3b1b
A dermatologist's take (Matt Zirwas), well written and reasonably sensible: some sun is probably good, too much avoidance may be bad, too much sun is also bad, and if you are outside for long stretches, wear sunscreen:
All of which leaves us with a mess of incomplete data and healthcare wrapped in politics, on a question none of us can opt out of (even those of us who want to hide under a rock, out of the sun).
AI Impact
AI Influencer Nav Toor discusses the recent retraction of a Nature paper that (supposedly) demonstrated that ChatGPT is a valuable tool in aiding student learning.
The retraction was quiet. However, the paper was viral. Over 500 papers already cite this study. Those citations are now embedded in other research.
Meanwhile, the studies showing that AI reduces critical thinking, creates cognitive dependency, and degrades skills when removed have not been retracted. They keep replicating. Across MIT, Stanford, Zurich, and The Lancet.
Retraction statement from Nature:
The initial paper got retracted quietly. However, the paper’s conclusions will likely continue to propagate, cited by papers and business white papers whose readers may never know the retraction happened.
A JAMA AI time-motion study followed 10 emergency physicians at UC San Diego during shifts with and without an AI-enabled medical record summarization tool. When an LLM summarized the chart, physicians saved all of 1.2 minutes, and only six of ten sped up at all. However, the participating physicians felt that the system was helpful and valuable.
I think this is part and parcel of what we're going to see with these AI tools - no single tool is going to be a home run winner in terms of objective measurements of improved outcomes. However, improving clinician quality of life, the perception of work efficiency, and, for a subset of clinicians, outcomes will likely be the norm. It's going to take a layering of these tools to make that happen.
JAMA Article:
AI-Assisted summary:
The J&E Random Kidney Facts of the Week
Two loyal readers recently suggested I salt the newsletter with random kidney facts for the nephrologic edification of the readers. Be edified, friends!
As an embryo, your kidneys begin low in the pelvis and climb to their final position (in the low back) during fetal development. The two kidneys ascend separately, each up its own side (like little rock climbers), and most of the time, arrive where they belong. Sometimes the two kidneys fuse into a single U-shaped kidney (called a "horseshoe kidney") that snags partway up and never finishes the trip. Sometimes one kidney stays in the pelvis, where it may work fine or may not work at all.
More common than any of these development issues is a problem not with where the kidney ends up but with how its drain (the ureter) develops. The ureter, the tube running from each kidney to the bladder, is supposed to enter the bladder wall at a shallow angle, so that when the bladder squeezes, the wall pinches the tube shut and urine only travels one way (out the urethra and out of the body). When that angle is off, a fault present from birth, urine flows backward up the ureter every time the bladder contracts, a condition called vesicoureteral reflux. Over time, that backwash (and the infections it invites) can quietly scar the kidney.
Reflux often goes unnoticed until someone is in their teens or adulthood, when one small, damaged kidney is discovered incidentally - the one remaining good kidney can often do enough cleaning and biochemical balancing that lab data does not show any diminished function.
Things I learned this week
NPR covered the very non-egalitarian diets of wealthy colonists who set the American Revolution in motion. This piece covers the culinary choices of Washington, Jefferson, and William Paca (early Maryland Governor and signer of the Declaration of Independence). Self-evident truths did not include that all menus are created equal. And I have mixed feelings now knowing our founding fathers ate stewed kidneys. (If you come visit me, I’ll take you to the Annapolitian establishments mentioned in this story, but will not serve you any organ I help care for.)
In 2024, researchers and religious leaders installed an artificial-intelligence Jesus, based on ChatGPT 4.0 at the time, to engage with parishioners and run it as both an art installation and a research project, observing how people responded to engaging with a pseudo-savior.
This week, I learned that some Hindu temples are starting to install robotic elephants to allow the public to interact with the venerated creatures without the risks, burdens, or concerns associated with animal mistreatment. In fact, one engineer highlighted in the article, Prasanth Prakashan, seems to have a robust business in mechanical pachyderm manufacturing.
AP Article
Meet Prasanth:
Artificial cells, robotic elephants, and AI Jesus all feel very deus ex machina.
AI art of the week
A visual mashup of topics from the newsletter and an exercise to see how various LLMs interpret the prompt. I use an LLM to summarize the newsletter, suggest prompts, and generate images with different LLMs.
A visual mashup of topics from the newsletter and an exercise to see how various LLMs interpret the prompt. I use an LLM to summarize the newsletter, suggest prompts, and generate images with different LLMs.
A Byzantine devotional icon painted in egg tempera on a gessoed wood panel, in the manner of a 12th-century master of Constantinople. Flat burnished gold-leaf ground covering the entire background, tooled with a fine punched halo pattern. Hieratic frontal composition, inverse perspective, no naturalistic depth or cast shadows. Muted earth palette of ochre, terre verte underpainting showing through the flesh, deep vermilion and lapis robes, with fine gold striations (chrysography) on every garment. Solemn, elongated figures with long noses and small mouths, gazing directly outward. Faint craquelure and worn gilding at the edges, as if venerated for centuries.
At the center, enthroned in the position of Christ Pantokrator, sits a robotic elephant rendered as a sacred figure: brass-and-enamel plating worked like cloisonné, one mechanical trunk raised in a gesture of blessing, the other holding a closed gilt gospel book. A halo of hammered gold encircles its head, inscribed with faux-Greek lettering.
Flanking the throne in the positions of attending saints stand two figures. On the left, a haloed figure of Christ whose face is a softly glowing screen, robed in traditional red and blue, one hand raised in benediction, the other holding a scroll of shimmering text. On the right, a tonsured physician-monk in a deacon's robe, holding not a censer but a small glowing tablet, its light falling on his face like divine radiance.
In the lower register, presented as venerated relics on a draped altar cloth, sit three small, potato-shaped, translucent cells, each haloed in gold and dividing into daughter cells, with a tiny painted spoon offered toward them as if feeding them. A hand-lettered titulus in mock-Greek majuscule runs beneath: the words "SPUDCELL" and a cross.
Around the border, a continuous decorative frame of interlaced gold vinework, with four small roundels in the corners containing: a coiled DNA helix, a sun disc crossed out (for sunscreen), a set of scales tipped by coins (for conflicted committees), and a withering laurel over a coin (for defunded science). The whole panel lit with the flat, even, sourceless glow of gold-ground icon painting. Reverent, solemn, and utterly deadpan.
Clean hands and sharp minds,
Adam
Comments
Post a Comment