Week of August 25, 2025
I've been 50 for four months. The metric of a successful week is now marked by getting new glasses, completing multi-staged dental work, and finding my favorite nutritional supplement in new travel-friendly packaging.
---
You can listen to a Google Notebook LM A.I.-generated podcast of the newsletter with two virtual "hosts."
This week, the hosts think I am a neuroscientist.
https://drive.google.com/file/d/1d2T2FEm-Dna7Cgwt2Z_2ptV1qwSd3auH/view?usp=sharing
------
Science and Technology Trends
Dr. Dominic Ng, a neuroscientist from the UK, shared an analysis of data from 185,012 UK Biobank participants followed over 10 years. The researchers employed machine learning techniques to identify optimal food combinations that reduce dementia risk more effectively than existing dietary patterns. Though the article is behind a paywall, a loyal reader shared a copy (thank you, Dr. B!). Dr. Ng's X thread summary is a fantastic commentary demonstrating how AI/ML makes it easier to understand many-to-many relationships in large data sets. It is important to remember that these kinds of precision nutrition studies are not causal, yet an excellent way to make and test further hypotheses and interventions. And, it is challenging for any one person to know how to implement such data in their own life, especially if their exact risk of dementia is unknown or the "optimal diet" for dementia conflicts with other health concerns.
https://x.com/drdominicng/status/1958137957023686862
and
Article (behind paywall):
https://www.nature.com/articles/s41562-025-02255-w
AI Summary of the original article:
https://claude.ai/public/artifacts/f530bc88-7c4a-4ebb-b151-b7722c24a166
Slate recently wrote about a JAMA cross-sectional analysis of Human Mortality Database data comparing US mortality to population-weighted averages of 21 other high-income countries between 1980 and 2023. These data demonstrate that, irrespective of the pandemic, US individuals under the age of 65 have a 20-30% increased death rate as compared to individuals in other countries. While the authors do not identify a single most significant factor, drug overdose deaths, firearm injuries, cardiometabolic diseases, Health system inadequacies, economic, social, and political determinants of health are likely contributors. The disproportionate impact on younger adults raises numerous national security and financial concerns. These excess deaths result in substantial years of life lost and economic productivity. [I had Claude add some analysis comparing this article to Upton Sinclair's The Jungle, so scroll to the bottom of the AI analysis to see some college-level discussion of unfettered capitalism.]
Slate Article (paywalled): https://slate.com/technology/2025/08/millennials-gen-z-death-rates-america-high.html
Non-paywalled:
https://www.smry.ai/proxy?url=https%3A%2F%2Fslate.com%2Ftechnology%2F2025%2F08%2Fmillennials-gen-z-death-rates-america-high.html
Original JAMA article:
AI Review of JAMA article:
https://claude.ai/public/artifacts/27134f34-3a0c-440f-b480-c6d7421290ef
Anti-Science Articles of Note
This week, the article is not anti-science, but the topic is. I care for a subset of patients who use small bits of data to extrapolate into generalizations - a combination of causation/association confusion and then demanding proof that they are wrong (the burden of proof fallacy) - "my friend got the COVID vaccine and died two days later, prove to me the vaccine didn't cause the death." I see several articles each year discussing electromagnetic field (EMF) radiation (Wifi, cell phones, etc.) Like the vaccine skeptics, I see comments like "my friend who had a brain tumor used their phone all the time, therefore cell phones must cause brain tumors." I recently found an article from a prolific physician epidemiologist (Mandier Setia, MD, PhD, MPH). The study was rigorous but flawed. Dr. Setia ran a prospective cohort study of 105 neonates in Navi Mumbai, India. His team investigated the association between radiofrequency electromagnetic field (RF-EMF) exposure (cell phones, wifi, etc.) and neurodevelopmental outcomes in the infants. The researchers directly measured RF-EMF levels in homes using professional equipment and tracked neurodevelopmental outcomes using the Ages and Stages Questionnaire (ASQ) over the first year of life. Even after controlling for birth weight, maternal age, and income level, the authors found several notable adverse outcomes in children whose homes had high degrees of EMF exposure - delays in fine motor skills, problem-solving ability, and social development.
However, the study included a small sample size of only 105 neonates over 1 year in a single location (Navi Mumbai, India). The researchers did not measure the timing or duration of RF-EMF exposure. Moreover, numerous confounders, such as maternal stress, environmental toxins, an unclear mechanism of action (biologically, what is happening?), or unclear threshold of exposure prior to seeing an impact (how much EMF? What bands of EMF? Is 1 year enough? Do these differences matter in the long-term? What does a “control” look like given the prevalence of all sorts of EMF in our world?) make this study hard to generalize. Lastly, there is very little other data relating typical human EMF exposure (on any band of the RF spectrum) to health outcomes.
My concern is that this type of article (flawed, but one that projects a reasonable, thoughtful degree of attempted scientific rigor) can justify hasty conclusions and fallacious action from people looking for data to justify their beliefs (confirmation bias). For example, one anti-science X account, GMOSciene, posted a link to this article with the comment "PARENTAL ALERT: What more proof do we need? Wifi harms our children, and here's an excellent peer-reviewed study." I note that the article looked at all forms of EMF (not just or even sub-segmenting EMF by source, such as wifi) and the open-source journal is of dubious peer review quality. These data are hardly some intellectual slam-dunk.
My takeaway is that this article does not clarify if, how much, and what types of RF radiation are associated with any sort of medical outcomes (what if all of the measured children grow up to be MD, PhD, MPHs? No long term follow up!). I have no reason to stop using my wifi, stop using my cell phone, or build a Faraday Cage around my house.
Here is the article:
https://www.cureus.com/articles/381425-radiofrequency-electromagnetic-field-emissions-and-neurodevelopmental-outcomes-in-infants-a-prospective-cohort-study
Here is an AI summary of the article:
https://claude.ai/public/artifacts/2e57ace5-dad8-4ef7-9987-b47c9ab70dec
Here is an AI-driven synopsis of the available data (and the quality of that data) on EMF exposure:
https://www.openevidence.com/ask/69b744f2-08dc-421b-85f9-ffecb753ab2b
Here is the US Department of Commerce RF frequency allocation chart (bands and bands of RF-EMF! We are surrounded!):
https://www.ntia.gov/page/united-states-frequency-allocation-chart
Here is some data from NIH on the impact of RF radiation (in very high doses) on mice:
https://www.niehs.nih.gov/health/topics/agents/cellphones
Lastly, I asked ChatGPT to help me brainstorm on retrofitting a 3500 square foot house with a Faraday Cage:
https://chatgpt.com/s/t_68ab4dd554208191a6384eb7b4804ec3
Living with AI.
"There are now $40 BILLION worth of US data centers under construction, up +400% since 2022. For the first time in history, the value of US data centers under construction will soon EXCEED office buildings."
https://x.com/kobeissiletter/status/1958572939579637800?s=61
ChatGPT 5 can now demonstrate novel solutions (proofs) to complex math problems. These threads on X are esoteric (for non-math folks like myself). Still, the point is clear - LLMs can extrapolate from human knowledge and develop novel ideas, concepts, or solutions in some highly technical subject matter.
https://x.com/vraserx/status/1958211800547074548?s=61
And
https://x.com/sebastienbubeck/status/1958198661139009862?s=61
Infographics
Here is a map showing where owning a kangaroo in the United States is legal. This thread is seven years old, and numerous individuals have commented on the map's accuracy (see the discussion). It is nice to see some bipartisan agreement on a topic, given that a relatively small number of both red states and blue states allow kangaroo husbandry without a permit.
https://www.reddit.com/r/MapPorn/s/wIlLPD9Yan
Things I learned this week
I was unaware of the history of atomic gardening - using radioactive materials to force genetic mutation in plants, looking for beneficial changes. Generating mutant plants is like the OG of GMO, right? (I wonder what the X user GMOScience [see above on EMF] would think?)
https://en.wikipedia.org/wiki/Atomic_gardening
And
https://www.nytimes.com/2007/08/28/science/28crop.html?smid=nytcore-ios-share&referringSource=articleShare
YOU NEED TO STOP WHAT YOU ARE DOING (like reading this newsletter) and click on the link below. I am shocked no one has previously alerted me to the "Dance your PhD" contest in which scientific researchers make music videos explaining their work (since 2008). Why did I not know about this?!? Collective Phenomena in Ultracold Dipolar Quantum Gases has never felt sexier.
https://www.science.org/content/article/winner-2025-dance-your-ph-d-contest
Two amusing headlines this week:
"Florida man arrested after crashing meth, fentanyl-filled drone into home and gets arrested after waking the home owners at 1 am to politely request access to the backyard [to retrieve said drone]." Lesson - good manners are essential, but you are better off trespassing if your drug-laden drone lands on private property.
https://ground.news/article/drone-carrying-drugs-crashes-into-lutz-home-suspect-arrested-after-knocking-on-door?utm_source=mobile-app&utm_medium=newsroom-share
"Zombie squirrels covered in oozing warts spotted skulking through US backyards: report." Sadly, there is an outbreak of squirrel fibromatosis — a viral skin disease caused by leporipoxvirus, which is not contagious to humans or other animals. If the afflicted squirrels could talk, I suspect they would take umbrage at the notion that they are "zombies" that are "skulking."
https://nypost.com/2025/08/17/us-news/zombie-squirrels-covered-in-oozing-warts-spotted-sulking-through-us-backyards/
AI art of the week
A visual mashup of topics from the newsletter.
I use ChatGPT to summarize the newsletter, suggest prompts, and make the images.
A 1980s 8-bit video game style image combining three storylines: Level 1: Scientists in lab coats dancing on a neon-lit stage with test tubes as microphones (Dance Your PhD). Level 2: A gardener in goggles tending glowing mutant vegetables under a giant atomic symbol (Atomic Gardening). Level 3: A split scene with a Florida man sneaking toward a house to retrieve a crashed drug-filled drone, while on the other side, zombie-like squirrels with cartoon warts skulk around. The whole image should look like a multi-level side-scrolling video game, pixelated and colorful.
Gemini https://drive.google.com/file/d/1VkIYxzR6fUzjEp5Cjwhh5yIKaFj2CaGr/view
Grok
https://drive.google.com/file/d/1mKIsWh8VzHaKvSGcbXktFS8e7Obu1qmK/view
ChatGPT
https://drive.google.com/file/d/1MJ5D8HqNYto_BdrFfQrzowzkJGSOYf_2/view
---
The Pandemic Mitigation Collaborative (PMC) utilizes wastewater viral RNA levels to forecast four-week predictions of COVID-19 rates.
Rates are going up - we are in a new wave.
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