Week of December 1, 2025
I appreciate the time and space around Thanksgiving; our kids are home for days, we have time for family games and fire-pit nights, and only one required event (which I am grateful is not at our house). At 50, traditions feel a little worn - previous years have faded into warm memories, softened by rosy retrospection, and the present sometimes can't compete. What I look forward to now is long conversations while watching wood burn and eating s'mores paired with smoky whiskey.
I didn't realize s'mores is a contraction of "some more," first appearing in the 1927 Tramping and Trailing with the Girl Scouts (a title that was probably just earnest and wholesome then, but now feels awkward and mildly inappropriate).
https://www.google.com/books/edition/Tramping_and_Trailing_with_the_Girl_Scou/M5zkAAAAMAAJ?hl=en&gbpv=1&printsec=frontcover&bsq=%22some%20more%22
---
Listen to a Google NotebookLM-generated podcast of the newsletter, featuring two virtual hosts.
https://drive.google.com/file/d/1wOW8JFz2VxAhscKs_uUZZ4Z8wEYIywH7/view
------
Science and Technology Trends
JAMA published a comprehensive review of the evidence (or lack thereof) for the various medicinal uses of cannabinoids. The review is an excellent forcing function for good clinical thinking - we have some known data (positive association between THC and cancer treatment-related nausea, for example), a lot of known unknowns (data suggesting some increased cardiovascular risk with THC use, but how much is unclear), and lots of unknown unknowns (are there long-term psychological or neurological effects?). Despite the gaps in knowledge, I have patients "successfully" managing pain and sleep disorders with THC-containing products from Maryland's dispensaries. In practice all this means the conversations I have with those patiets come back to, "you are using a substance that we don't fully understand the risks/benefits of and I will keep my eye on the literature to help your decision-making. In the interim, you may be experiencing benefits or the placebo effect, without full clarity as to the risk of use." (FYI - this is what the pharma ad's "talk to your doctor" might look like.)
Paper:
https://jamanetwork.com/journals/jama/fullarticle/2842072
AI-supported Analysis:
https://claude.ai/public/artifacts/160c6eec-9436-4890-a1fc-453b896aa213
A note on potential bias in the JAMA article (and an interesting use of AI): In light of recent federal legislation restricting THC-containing products, I asked Claude to examine relationships between the paper's authors, the legislation's timing, and organizations that supported the restrictions. Claude found connections to anti-cannabis advocacy groups that may broadly explain the article's timing.
About the federal legislation:
https://thehill.com/policy/healthcare/5606807-hemp-loophole-government-funding-bill/
Analysis of the paper's authors and relationships:
The BBC offered an excellent review of the UK's recent updated prostate cancer screening recommendations. The article nicely weaves together clinical epidemiology, politics, individualized risk tolerance, and anecdotal data - it is an excellent example of why operationalizing clinical recommendations across an entire population can be challenging and controversial. The new recommendations reflect that current mass screening results in roughly six men undergoing unnecessary biopsies or treatment for every one life saved. The BBC article captures the various trade-offs well. Of course, every circumstance is different. Clinicians should screen high-risk and other appropriate patient populations.
BBC Article:
https://www.bbc.com/news/articles/cm20169gz44o
Here is the article's sensitivity/specificity infographic:
https://ichef.bbci.co.uk/news/1024/cpsprodpb/f0d9/live/ed6fac30-cc89-11f0-a892-01d657345866.png.webp
AI-supported analysis:
https://claude.ai/public/artifacts/4acdd98e-5289-4443-92f5-4091ba1a781e
Another week, another pair of articles confirming my confirmation bias on the medical value of coffee.
Paper 1
Coffee may be associated with less cellular aging. Norwegian researchers performed a cross-sectional study of 436 mentally ill Norwegians, measuring part of their DNA called the telomere, compared to reported coffee consumption. Their data suggested that patients who consumed 3-4 cups of coffee daily had the longest telomeres (longer is better; telomeres shorten as cells age). The observed telomere lengths correspond to approximately 5 years younger "biological age" than non-coffee drinkers, after adjusting for confounders. [FYI - The choice to study patients with severe mental illness is not well explained and makes generalizability uncertain.]
Article:
https://mentalhealth.bmj.com/content/28/1/e301700
AI-supported analysis:
https://claude.ai/public/artifacts/dd74b9df-29b8-4e50-8022-5fae14a29b4e
Paper 2
Researchers used fMRI to examine the brains of coffee drinkers 30 minutes after consumption. Coffee quieted the brain's mind-wandering regions while boosting activity in areas governing vision and decision-making—essentially priming drinkers to focus. Interestingly, when participants drank caffeine dissolved in hot water instead of actual coffee, only the mind-wandering regions quieted. The decision-making boost didn't appear. Something about coffee itself (the ritual, the aroma, compounds beyond caffeine) seems to matter.
Article:
https://www.frontiersin.org/journals/behavioral-neuroscience/articles/10.3389/fnbeh.2023.1176382/full
AI-supported analysis:
https://claude.ai/public/artifacts/1e73bccd-9050-43a8-bfee-699c7a4bdf8a
FYI: In reviewing this paper, I learned that the Institute for Scientific Information on Coffee (ISIC) funded the research. ISIC is a non-profit backed by Illy, Lavazza, Peet's, and other major roasters (big coffee?). The studies they cite appear rigorous, but I now have another layer of bias to filter when reading coffee research. I hope ISIC doesn't turn out to be like: https://en.wikipedia.org/wiki/Tobacco_Institute
Anti-Science Articles of Note
Science Vs. is one of my favorite podcasts for data-driven discussions of science. Their 10th anniversary podcast is well worth the 40-minute time investment. The topics they covered included their reviews of:
- Vaccines and autism: They revisit Andrew Wakefield's flawed MMR–autism study, explain its retraction, and summarize extensive studies showing no link between MMR vaccination and autism.
- Misinformation and ivermectin: They recount a tense interview with Dr. Pierre Kory, who, despite excellent data to the contrary, strongly advocates ivermectin for COVID.
- Orgasm research: An interview with Nan Wise, a sexual health researcher, who performed a first-of-its-kind fMRI orgasm study. Dr. Wise describes the many practical problems of studying orgasms in an MRI. Here is her paper: https://pubmed.ncbi.nlm.nih.gov/28986148/. I suggest you review the paper's methods section, then listen to the podcast, noting the difference between the written "professionally described" methods section and what actually happened during the research.
- Paleo diets: Anthropologist Katharine Milton argues there was no single paleo diet and that proponents of "ancient diets" are not eating anything similar to anything Paleolithic humans ate.
https://open.spotify.com/episode/5bHaR022fkh6i0X8tQSxjg?si=4edCsx7-R2WK6dySTZZxfQ&t=654&pi=iEKIX-JOTYezP
Living with AI
Dr. Eric Horvitz, MD, PhD, Chief Scientific Officer at Microsoft, was interviewed on the JAMA learning network podcast. The 24-minute interview touched on one of my AI focus areas - the challenges AI tools pose for healthcare delivery, given their inherent inconsistency (LLMs like ChatGPT put out variable answers - i.e., non-deterministic, which is problematic when clinical care requires a relatively high degree of consistency). Striking the balance between efficiency and safety sums up so much of the struggle I see: tools that offer unprecedented capability to summarize, capture, characterize, and process voluminous data, and the strong need for technical and operational guardrails (to ensure the output is sound and that humans still add judgment and context).
Podcast: https://edhub.ama-assn.org/jn-learning/audio-player/19019635
AI-supported summary of transcript: https://claude.ai/public/artifacts/c83a672a-08d4-47f1-af60-1b99d3dc2435
AI is also accelerating bioengineering. I'm curious how close we are to John von Neumann's notion of exploring our galaxy via "[biological] self-replicating automata." In other words, replace conventional electronic spacecraft with living, engineered microorganisms designed to survive space travel and phone home (in a manner of speaking).
Excellent X discussion that reminded me of von Neumann's ideas and the link to biology:
https://x.com/nikomccarty/status/1994843829170765916
An interesting linked (speculative futurist) paper exploring "the challenges of a picogram-to-nanogram-scale probes [sent throughout our galaxy on solar sails] to land, replicate, and produce a communications module based on biominerals at the destination."
Paper: https://journals.sagepub.com/doi/10.1089/ast.2022.0008
AI-supported analysis:
https://claude.ai/public/artifacts/d17a2140-f4dc-4e0e-9aed-101d66bcde5a
Consider how AI-supported research can drive bioengineering. "Since Google released AlphaFold in 2020, the AI [protein structure tool] has helped researchers all over the world to predict the 3D structures of hundreds of millions of proteins."
https://www.nature.com/articles/d41586-025-03886-9
Infographics
I am still playing with Gemini 3. Here is an infographic generated from the content included with the bioengineered Von Neumann probes. It is amazing how little effort it takes to generate high-quality images that take a wide range of source data, synthesize it, and put the key elements into a visual representation:
https://drive.google.com/file/d/1uiuLWXj5Law2H7sPxWJ8cUhPbYEJljln/view
Things I learned this week
A fascinating intellectual rabbit hole marked our Thanksgiving dinner. Several attendees discussed their experiences with fortune tellers and tarot card readers. I began wondering whether tarot card reading requires education and certification. I was surprised (and not) to learn online educational platforms offer a wide range of "esoteric practice" course studies, including tarot reading, Reiki therapy, conjuring, and (my personal favorite) "psychic self defense" (for, I learned, "defending against psychic attacks and energy vampires"). While determining value is difficult, the $8 to $20 price tags on these courses feel like (at worst) low-cost entertainment. I wonder whether a 'Udemy-certified' label inspires confidence in Reiki clientele, or whether that's precisely when one might want the psychic self-defense course.
See all the offerings:
https://www.udemy.com/course/beginners-hoodoo-conjure/
Psychic Self-Defense is now 85% off at $9.99!
https://www.udemy.com/course/psychic-self-defense-mastering-energy-protection/
Sadly, my older son's brilliant idea for the artificial tongue was thought of by others. However, this tech will need scaling and product development to realize his notions of teletasting and related remote gustatory activities. "Scientists in China have developed a soft and flexible artificial tongue that can detect the heat level in spicy foods, and support or reduce reliance on human taste panels. The device can detect capsaicin and pungent-flavored compounds found in garlic, ginger, and black pepper, which could streamline spiciness testing."
https://www.foodingredientsfirst.com/news/artificial-tongue-spiciness-testing.html
https://www.nature.com/articles/d41586-025-03767-1
and
Article: https://pubs.acs.org/doi/abs/10.1021/acssensors.5c01329
If the 16th-century Dutch sent artificial tongues for remote-tasting to Mauritius, would the world still have Dodo birds?
"The very first written account of the dodo was recorded by Vice-Admiral Wybrand van Warwijck, with the first fleet of Dutch ships to reach Mauritius in September 1598: We called these birds Walghvogels (disgusting birds), partly because, [despite stewing them] for a very long time, they were tough to eat, [oily, and not tasty]."
https://www.sfchronicle.com/food/article/The-way-of-the-dodo-A-recipe-for-disaster-13885936.php
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 vintage 1930s Works Progress Administration (WPA) National Park poster, rendered in a bold screen-print style with thick lines, geometric shapes, and a limited palette of muted earth tones, forest greens, deep blues, and electric yellow. The central figure is a heroic, stylized 17th-century Dutch sailor with a wide-brimmed hat, kneeling before a blazing campfire and holding a perfectly roasted marshmallow on a stick, ready to make a s'more. A radiating, geometric psychic forcefield composed of yellow and blue light beams protects the campsite. Crashing uselessly against the outside of this forcefield are shadowy, translucent "Energy Vampires"—stylized figures in 1930s business suits holding briefcases, recoiling in defeat. Safely inside the protective barrier, the sailor sits with a sleek, futuristic artificial tongue device on a brass stand, its display glowing, next to a blocky, stylized Dodo bird observing the fire. The background is a rugged, stylized coastline with pine trees. Bold, blocky typography across the top reads: "FORTIFY YOUR MIND & YOUR MARSHMALLOWS." At the bottom: "CERTIFIED PSYCHIC DEFENSE. U.S. DEPARTMENT OF THE INTERIOR." The image has a textured, serigraph ink-on-paper feel.
I think Google's Gemini 3/Nano Banana Pro image generator wins again this week.
ChatGPT: https://drive.google.com/file/d/1RyyHaEFQ6OgZRe7yXOMFME8EC-p2i5QO/view
Perplexity:
https://drive.google.com/file/d/1ZCWxx7Lhtr-ljlc-GrHmoRyIm8-P-DTE/view
Google Gemini
https://drive.google.com/file/d/1xt-1UOFdIJ9wdU05PXWUrGHAdJ0rHPrq/view
Grok
https://drive.google.com/file/d/1Q7M3pp6NHlMGF0CiMVpB0Xl_uDDCaVaR/view
---
Recall that this newsletter began with the pandemic.
COVID rates are trending upward, likely due to holiday travel and gatherings.
The Pandemic Mitigation Collaborative (PMC) uses wastewater viral RNA levels to forecast COVID-19 rates over the next 4 weeks.
https://pmc19.com/data/
based upon https://biobot.io/data/
Wastewater Scan offers a multi-organism wastewater dashboard with an excellent visual display of individual treatment plant-level data.
https://data.wastewaterscan.org/
----
Clean hands and sharp minds,
Adam
Comments
Post a Comment