Zettelnaut
The Series Where I Follow Connections the Research Hasn't Delineated Yet
In August 2023, I wrote a note on a study published in the New England Journal of Medicine. Kooshesh and colleagues at Mass General had found that adults whose thymus was removed during cardiac surgery had nearly triple the five-year all-cause mortality of matched controls. The thymus, an organ most physicians consider vestigial by adulthood, apparently wasn’t vestigial after all.
That note sat in my system for months before it collided with something unexpected.
Over the next two and a half years, I wrote notes on caloric restriction regrowing the human thymus on MRI (Science, 2022). Exhausted T-cell populations in Long COVID (Nature Immunology, 2024). Antimicrobial resistance projections through 2050 (Lancet, 2024). Zonulin-mediated gut permeability in post-COVID immune dysregulation (Science Translational Medicine, 2025). An AI-derived thymic health score that independently predicts cancer immunotherapy response (Nature, 2026). Nine notes across four years. None of the source papers cited each other. Their authors are at different institutions and read different literature. A paper on thymic involution does not cite a paper on butyrate-producing gut bacteria, even when both describe the same inflammatory cascade.
My Zettelkasten linked them.
The thymus turned out to be the upstream regulator connecting immune aging, cancer treatment, post-viral syndromes, gut barrier integrity, and the coming antibiotic resistance crisis. No individual paper contained that argument. No review article had assembled it. The connection existed in the literature, distributed across journals and years. That became an essay called “The Lifeline.”
This is what a Zettelkasten does. I’ve maintained one for over a decade, a personal knowledge system where every note links to other notes by concept rather than by discipline. It now holds hundreds of notes drawn from neuroscience, microbiology, exercise physiology, nutrition, aging research, and clinical medicine. Each note captures a single finding from a single paper, linked to whatever it connects to, regardless of field. Occasionally, the system surfaces an unexpected connection. A note from one reading session links to a note from a completely different field, written months or years earlier. The original papers never cite each other. But the mechanism underneath is the same.
Penicillin resulted from a contamination event that was noticed by someone working in microbiology and immunology. The link between H. pylori and gastric ulcers required a pathologist willing to ignore gastroenterological dogma. The structure of DNA required X-ray crystallography, organic chemistry, and genetics to work together. The pattern holds: the connection between fields is where new understanding forms. It almost never comes from going deeper into only one.
Zettelnaut is the series where I follow those connections. Each essay traces a single thread through multiple research papers that don’t cite each other, identifies the shared mechanism, and follows the implications.
These essays were built using an AI-augmented Zettelkasten. I'm building a course on the method. Get early access here:
Microplastics Are the New Lead Paint
April 26, 2026
Brain tissue from 2024 decedents contains roughly 7 grams of plastic — about a plastic spoon's worth — and the concentration was 50% higher than in samples taken eight years earlier. The primary delivery vehicle is the food supply: ultra-processed foods carry microplastics, BPA, and phthalates from packaging and processing equipment directly into the body, where the route from gut to brain runs through a barrier system already compromised by the Western diet that delivers them. The structural parallels with lead paint are not metaphorical — ubiquitous environmental toxin, consumer-product delivery, decades of latency, disproportionate impact on vulnerable populations, industry incentive to downplay. This essay traces the gut-brain-vascular pathway connecting diet to tissue accumulation and argues that we are repeating the regulatory delay that defined the lead-paint era — while honestly engaging with the January 2026 methodological controversy that complicates precise numbers without changing the direction of the signal.
Life’s Final Draft
April 15, 2026
We prepare for death medically and materially. We have no practice in preparing ourselves. What follows is a clinical thought experiment called Life’s Final Draft: an AI trained on a dying person’s behavioral record, used not to comfort survivors but to help the dying person confront themselves. The twist is that the same process produces a vetted, consent-based deathbot for the family to use after death. The curation (what to keep, what to remove, what to say that was never said) IS the existential reckoning. The risk is that an AI in your voice could become the most sophisticated self-justification machine ever built. The design has to be adversarial, not validating. Socratic, not sycophantic. Otherwise, it’s denial technology in a dying person’s hands.
Don’t Ban the Grief Bot. Prescribe It.
April 14, 2026
A fourteen-year-old dies after months with an AI companion. Replika users hold digital funerals for their “lobotomized” chatbots. The instinct is to ban the technology. But what if that’s the wrong question? The same tool that destabilizes when unsupervised may be therapeutic inside a clinical frame — because the molecule didn’t change with psychedelics, the container did. Modern bereavement science has moved on from closure: continuing bonds theory holds that healthy grief transforms the relationship, not severs it. A chatbot trained on a dead person’s words isn’t necessarily pathological attachment. It might be a transitional object that talks back. The 2025 “Artificial Continuing Bonds” framework shows what responsible use looks like: informed consent, defined boundaries, clinical oversight — a guided session, not an unsupervised trip. People are already talking to AI versions of the dead. The question is whether clinicians engage or cede the territory to tech companies whose incentive is engagement, not healing.
The Lifeline
March 31, 2026
Kooshesh and colleagues at Mass General followed 1,420 adults who had their thymus removed during cardiac surgery. At five years: 2.9× higher all-cause mortality. Double the cancer risk. In March 2026, Bernatz et al. published in Nature: a deep-learning model trained on routine chest CTs can derive a thymic health score, and that score independently predicts immunotherapy response with effect sizes in the same range as PD-L1 and tumor mutational burden. In 2022, Dixit’s group at Yale showed that two years of moderate caloric restriction measurably regrew the human thymus on MRI. The organ medicine stopped thinking about may determine whether your immune system can still defend you.
The Cooling
March 25, 2026
Carl Wunderlich set the body temperature at 98.6°F in 1868. Julie Parsonnet’s group at Stanford analyzed 677,000 measurements spanning 157 years and found a steady decline of about 0.03°C per decade of birth. The average American now runs closer to 97.5°F. Bongers and Dickson’s group at Michigan showed that germ-free mice run significantly cooler than colonized controls, and identified Lachnospiraceae as the bacterial family most strongly associated with temperature trajectory. Over the same 150 years, industrialized populations lost an estimated 30 to 40 percent of their gut microbial species. The temperature and diversity declines are parallel curves plotted over the same timeframe.
The Munchies You Didn’t Smoke For
March 17, 2026
Erin Hanlon’s team at UChicago sleep-restricted healthy adults and watched their endocannabinoid 2-AG spike by day four, driving cravings for energy-dense food they didn’t need. The same CB1 receptor system produces the runner’s high. Raichlen’s group confirmed in 2012 that it’s endocannabinoids, not endorphins, that mediate the mood elevation runners describe. Dohnalová and colleagues showed in Nature that gut bacteria manufacture the fatty acid amides that make exercise feel rewarding in the first place. Sleep loss activates the same receptors that should reward running but instead point them at the refrigerator, while the ultra-processed food you eat when sleep-deprived destroys the gut bacteria that would make exercise feel worth doing.
The Two Gates of Sleep
March 10, 2026
A 2025 paper from Gero Miesenböck’s lab at Oxford found that sleep pressure is triggered by mitochondrial electron damage in specific sleep-control neurons. A separate 2025 study detected bacterial cell-wall fragments in mouse brains, fluctuating with the sleep-wake cycle, reviving James Krueger’s forty-year-old finding that microbial molecules can act as somnogenic signals. Together, they suggest sleep may require two gates: a biological readiness signal built from mitochondria and microbes, and a cognitive release mechanism that lets the mind stand down.
Each Zettelnaut essay connects research that doesn’t know it’s connected. If you want to see where the next set of notes leads, subscribe to HealthHippieMD.
These essays were built using an AI-augmented Zettelkasten. Want to build your own? I’m assembling a course on the method. Get early access here: