Week of August 18. 2025
I came home to find our college-age sons and their grandmother watching the 1967 film The Graduate. I enjoyed seeing multiple generations of my family bonding over an anti-counter-culture movie portraying adults as vapid, shallow, and materialistic. Listening to the three of them discuss "the significance of the film's diegetic vs. non-diegetic use of Simon and Garfunkel music" made me feel vapid and shallow. Of course, it's probably better to be like the film's superficial parents than Mrs. Robinson.
If, like me, you need a refresher from your college film course, my son was kind enough to “remind” me that diegetic sound refers to music and effects that the characters can hear and respond to, while only the audience hears non-diegetic sound (like musical scores). This concept relates loosely to the idea of the "fourth wall," which separates the characters' reality from the viewers'.
---
I did not have time to generate the podcast this week.
------
Science and Technology Trends
AstraZeneca now offers a home delivery program for Flumist, the only nasal spray flu vaccine in the U.S. It is a live, attenuated vaccine, with less efficacy than injectables, but is FDA approved for self-administration for adults aged 18-49 and caregiver-administration to children aged 2-17. It is not ideal for immunosuppressed patients, but it may be a reasonable option for home-bound or needle-phobic individuals. This program is the first time I am aware of a direct-to-consumer vaccine.
https://www.astrazeneca-us.com/media/press-releases/2025/FLUMIST-the-nations-only-nasal-spray-flu-vaccine-now-available-for-home-delivery.html
While I am thinking about vaccines, I found a recently published case-control study looking at tens of thousands of patients diagnosed with COVID. Researchers analyzed the Kaiser Permanente Southern California system data set, looking at the relationship between the number of COVID vaccinations and/or COVID treatments (remdesivir, nirmatrelvir/ritonavir) and the persistence of Long COVID symptoms after 6 months. While this is a retrospective analysis, the more exposure to the COVID vaccine and/or treatment with antivirals significantly reduced the likelihood of long COVID symptoms 6 months post-infection.
Article:
https://www.sciencedirect.com/science/article/pii/S221133552500227X
AI Summary:
https://claude.ai/public/artifacts/3d8daa7a-86f2-4338-9f72-96faa957e4e8
OpenAI is investing in Merge Labs, a competitor to Neuralink (Elon Musk's computer-brain interface company). I will let you ponder the notion of having ChatGPT engaged with humans via computer-brain interfaces, which may or may not be better than interfacing with Grok (which is what Neuralink will eventually do).
https://techcrunch.com/2025/08/12/sam-altman-openai-will-reportedly-back-a-startup-that-takes-on-musks-neuralink/
and
https://www.engadget.com/ai/openai-and-sam-altman-are-reportedly-creating-a-startup-rival-to-elon-musks-neuralink-123022874.html
and AI Summary:
https://www.perplexity.ai/search/summarize-lPebTQqcSO.9sz2oBHVwUg#0
Anti-Anti-Science Articles of Note
Dr. Sandro Gelea, from Boston University, wrote a fantastic blog post about "the value of engaging with difficult ideas from disagreeable sources," exploring the concepts of intellectual bias.
https://substack.com/home/post/p-170943019
AI Summary:
https://claude.ai/public/artifacts/616f9191-5ba5-49a7-9493-0515f2d0a2fe
Dr. Thomas Farley offered a thoughtful analysis of the supplement industry. This topic comes up frequently amongst my patients. Given the sheer number of supplements, the potential for interactions with other drugs, and the paucity of evidence, I often advise patients to avoid over-the-counter medicines without a clear, measurable goal and monitoring plan.
https://open.substack.com/pub/healthscaping/p/what-are-those-dietary-supplements
AI Summary: https://claude.ai/public/artifacts/beb33180-4811-4ce1-baf0-013fe928cbb3
AI has made looking for drug-drug interactions a lot easier. OpenEvidence.com (for clinicians) makes summarizing large bodies of evidence and applying that evidence to clinical questions much, much easier. (See the living with AI section for more info on OpenEvidence). Here is a prompt along with a framework I often use when thinking about supplements: "Give me a table of the 20 best studies on Vitamin D used as a supplement. Please list the study, the population size, the duration, the measured impact, side effects, and outcome. Based on these data, please suggest whether a 54-year-old patient with a VitD 25OH level of 18, hypertension, but otherwise healthy on only losartan, should use Vit D and how much. Please frame that answer in: what is the goal, is the goal measurable, what is the dose, what kind of monitoring, what are the trade-offs, and what side effects might the individual experience?"
https://www.openevidence.com/ask/e4f0cba1-f4a6-4031-82fb-05919d1ae9c6
Living with AI.
A good reminder that AI amplifies our abilities to think critically (or not). Here is a medical case report about an LLM-using individual who "conducted a personal experiment" for 3 months after reading about the negative health effects of sodium chloride (table salt). The individual consulted ChatGPT about chloride alternatives (he was reportedly disappointed that he could only find information on reducing sodium portion of NaCL from one's diet). The AI told him that "chloride can be swapped with bromide," whereby he purchased sodium bromide from the internet, and replaced his dietary sodium chloride with sodium bromide. He ended up in the ER with paranoid delusions (that his neighbor was poisoning him) and was diagnosed with bromism—a rare but reversible overdose caused by bromide ingestion.
Article:
https://www.medpagetoday.com/emergencymedicine/emergencymedicine/116975?xid=nl_mpt_DHE_2025-08-13
Journal Article: https://www.acpjournals.org/doi/epdf/10.7326/aimcc.2024.1260
AI Summary: https://claude.ai/public/artifacts/55b88a62-6ee3-4476-b447-c6591b5c5ef3
Related - articles on the impact of AI on education – are users likely to become dependent and less informed with A!?
https://open.spotify.com/episode/4ZHxqcy4YqmISQmoqaFVLJ?si=O3GuZayRQXC2XS6eG7_GFw
and
https://mndaily.com/294949/opinion/op-ed-were-losing-our-love-of-learning-and-ai-is-to-blame/
and
https://www.nytimes.com/2025/07/18/opinion/ai-chatgpt-school.html?unlocked_article_code=1.e08.c_Bp.jIcuNaefuoTg&smid=url-share
Bearing in mind that medical licensing exams and medical diagnosis are only part of the practice of medicine, OpenEvidence, an artificial intelligence startup (that I use frequently), has developed an AI model that scored a perfect 100% on the United States Medical Licensing Examination (USMLE). This achievement marks a significant leap in AI's capability to interpret medical information. It is a good reminder that the many-to-many relationship between clinical signs, symptoms, data, and diagnosis is easily tackleable by AI. Helping our fellow humans make good choices and nudge behavior over time is the real work of healthcare.
https://www.fiercehealthcare.com/ai-and-machine-learning/openevidence-ai-scores-100-usmle-company-offers-free-explanation-model
AI Summary: https://www.perplexity.ai/search/summarize-this-lUOVn6zAQYqPqM6aIahwYg?0=d#0
https://www.statnews.com/2025/08/12/ai-deskilling-doctors-colonoscopy-study-lancet/
https://claude.ai/public/artifacts/634aed9e-3829-4480-9ae2-47ca93f4f926
Infographics
Egg yolk color variations and causes. I have never seen a green yolk, but it is good to know they are "safe to eat."
https://hilltoholler.com/wordpress/wp-content/uploads/2023/10/Egg-Yolk-Color-Chart-HilltoHoller-1024x791.jpg
from:
https://hilltoholler.com/2023/info/egg-yolk-colors/
Things I learned this week
I learned that Brown University library has George Orwell's 1984 manuscript. I did not know that Orwell typically destroyed his manuscripts, but so ill with tuberculosis at the time of writing 1984, he was unable to do so.
https://www.brown.edu/news/2025-08-11/george-orwell-1984-manuscript
And for all y'all that are endlessly fascinated by words and writing, I found myself reading Yale's Grammatical Diversity Project for North American English. I so learned formal names for common English usage variations, of which the 'Drama So' is so my favorite.
https://ygdp.yale.edu/phenomena/alls-construction#who-says-this
and
https://ygdp.yale.edu/phenomena/drama-so
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.
Prompt:
In a cozy 1960s-style living room, a grandmother, two college kids, and a parent float gently above the sofa like balloons while The Graduate plays on the TV. Simon & Garfunkel peek out from the screen, tossing oversized musical notes into the room. At the front door, a fairy godmother–like postal worker delivers a glowing nasal spray box, while neighborhood cats and dogs line up politely for their turn. Outside the window, two bumbling robots shaped like old radios tug at a giant glowing brain balloon, which floats higher with every thought bubble they drop.
Style: Mid-century Whimsical Cartoon Modernism meets 1930s Silly Symphony animation, with touches of Dr. Seuss Expressionism.
ChatGPT: https://drive.google.com/file/d/1HLi7rvaRul4OxPpJOomPOZvt_yYYUnlV/view
Grok: https://drive.google.com/file/d/1pXbYVClBVaIVPY1O8W6YOYDnxvp57wzL/view
Gemini: https://drive.google.com/file/d/14bAWZ-YsOMUdUdy0IliqKKa1StR5wNTj/view
---
The Pandemic Mitigation Collaborative (PMC) utilizes wastewater viral RNA levels to forecast four-week predictions of COVID-19 rates.
1 in 93 people is infected (in other words, rates are increased again last week).
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