Week of February 16, 2026
Dr. Murphy, my long-time dentist, recently retired. During a cleaning last week, I met one of the new practice owners for the second time. As she manipulated my tongue (for the oral cancer screening), two thoughts struck me: I still don't like the taste of gauze, and I have only ever seen this woman upside down.
At this office, the dentist approaches from behind the headrest while the hygienist cleans. So in our two-visit, six-month relationship, my new dentist exists entirely as an inverted face—an upside-down bobbing chin admonishing me about brushing and flossing like a bad medical TikTok.
Of course, she probably sees me the same way—another upside-down mouth with 50-year-old teeth subjected to too much coffee and tea. Perspective can be so relative.
The Google Notebook LM AI-generated podcast version of this week's newsletter.
Science and Technology Trends
Singaporean government researchers released sterile male mosquitoes twice weekly for 24 months into various areas, monitoring mosquito populations and the incidence of Dengue fever. Thanks to sterile males competing with wild male mosquitoes, their data demonstrated both fewer mosquitoes and a lower incidence of Dengue fever. The study is government-funded research reported by the government agency responsible for Dengue control (i.e., no independent review), but I'm still amazed we can reshape infectious disease patterns with clever science. Also, I briefly imagined the human equivalent, a bad sci-fi movie about the scaled release of neutered Timothée Chalamet clones to college campuses, but suspect that is better left to another writer.
X-Discussion
Article
AI-supported analysis:
Last week's doctor-email clickbait rabbit hole led me to tattoo-related uveitis, a rare reaction in which tattoo ink triggers irritation and inflammation at the tattoo site and the middle layer of the eye. While this is rare (the review article cited 33 studies with 44 total patients published between 1969 and 2025), it raises several interesting issues:
1) It is a diagnosis of exclusion; to make the diagnosis, other autoimmune systemic diseases and infections need to be ruled out.
2) The mechanism is unclear - is it a typical allergic reaction (to an embedded allergen)? A triggered autoimmune disease? Something else?
3) This is a fantastic example of why being a thorough clinician is important - tattoos are not often asked about, and for the vast, vast majority of those who have tattoos, they never become an issue. Think about how many visits it must take to make this diagnosis. (Will AI even ask about tattoos?)
The clickbait article link:
AI-supported analysis (with some more background):
Adam! WTF! What is a uvea?
The rabbit hole was deep. I ended up searching for updates on e-tattoos, including embedded biosensors, microneedle patches, nanomaterial electrodes, and temporary tattoo sensors that can measure various biological signals, such as heart rhythms, chemical signals, and body fluid composition. To be clear, this is technology that is real, but still immature. The vast majority of devices are workable proofs of concept, not validated medical devices.
AI-supported analysis
Adam's Coffee Confirmation Bias Corner - February 2026 edition
(part of my ongoing search for data justifying my coffee, tea, and caffeine consumption).
JAMA published research using data from the Nurses' Health Study and Health Professionals Follow-up Study, which followed 131,000 health professionals for 43(!) years. Drinking 2-3 cups of caffeinated coffee or 1-2 cups of caffeinated tea daily was associated with an 18% lower risk of dementia. Decaf showed no benefit, suggesting caffeine may be protective. Drinking more than 2-3 cups didn't provide additional benefits, and the association held regardless of genetic risk factors.
Article:
AI-supported analysis:
Related: Older Adults in Japan who own dogs also have lower rates of dementia. It is unclear whether there is synergy between older Japanese adults and/or their dogs when caffeine is administered.
Anti-Anti-Science
Last week, the FDA refused to evaluate Moderna's mRNA flu vaccine despite positive trial data on efficacy and safety. Moderna had run the trials under a framework the Biden administration's FDA had agreed to. But the current FDA leadership (Dr. Vinay Prasad, Chief Medical Officer) found the control groups inadequate and declined review, overruling the FDA vaccine lead who supported the trial design.
I'm struggling to separate the issues here. It is the FDA's prerogative to set regulatory standards. And this feels like the goalposts are moving under the guise of letting 'perfect be the enemy of the good' (the Nirvana logical fallacy). The confusion may be the point.
Either way, this regulatory inconsistency undermines biotech investment.
STAT News Article:
X Thread by the STAT news article author, which goes through the Moderna data, which is promising.
AI Impact
Reuters published data indicating higher adverse-event reporting for AI-enabled surgical software (primarily AI-enabled video overlays for intraoperative clinical decision support). In one instance, at least 10 people were injured between late 2021 and November 2025, when J&J's TruDi Navigation System (an intraoperative video scope system for sinus surgery) misinformed surgeons about the location of their instruments while they were using them inside patients' heads. J&J is not alone - I have seen several articles exploring how post-launch FDA adverse event reporting for AI tools exceeds the historical patterns seen for non-AI-enabled tools. All of this is a good reminder that AI-enabled clinical tools require a higher threshold of testing, monitoring, and human-in-the-loop control.
Related: discussion of a September 2025 JAMA article looking at recall rates of AI-enabled FDA-approved devices and tools
(Adam observation: the irony is that AI-enabled surgical tools seem to face less scrutiny than control groups in mRNA vaccine trials.)
I found a fascinating (and unsettling) website called "The AI 2027 Scenario," the first major release from the AI Futures Project, a nonprofit focused on AI future forecasting composed of AI safety researchers, many of whom worked in industry.
I asked Perplexity what it thought of the future predictions (essentially, AI will achieve artificial general intelligence (AGI) and take over human society in the next few years, and that will be catastrophic, or we will figure out how to control it). Perplexity's analysis:
This is serious, well-researched work by credentialed forecasters aimed at quantifying AGI risks. It's more rigorous than typical AI doom scenarios, with explicit models and active error correction. However, the scenario format inherently emphasizes vivid specific outcomes over probability distributions, and the aggressive timeline (even if "modal") deserves scrutiny. Worth engaging with critically but not accepting uncritically.
I wonder what bias Perplexity brings to forecasting its own rise to power?
Things I learned this week
The New Yorker published a fascinating look at New York City's rat mitigation efforts, solving an evolving problem that emerged during the pandemic (when rats migrated away from closed restaurants and towards houses). The article focuses on Bobby Corrigan, a rodentologist who consults New Yorkers with rat-related infestations. Corrigan has clearly grown to admire rodents. This article may be the most pro-rat writing I have seen recently:
While working on a rat survey in a subway tunnel on the Lower East Side, Corrigan saw a rat in the beam of his flashlight. It didn't scamper away. Instead, it stood up on its haunches and stared directly at him. Most rats will run if you make a sudden movement. "This one did not," he said. "This rat was curious about me." Corrigan felt compelled to follow the rat. "I was, like, 'We had this moment. Is there anything to that?' "
I learned that Australia has the most genetically diverse feral camel herds in the entire world. Apparently, camels were brought to Australia for exploration of the western deserts, and now roam freely. In fact, some of these feral camels are rounded up and, via air transport, hundreds are sent each year from Australia to the Middle East for breeding, racing, and consumption.
Video of camels on cargo planes. Business class for sure.
In case you are thinking of importing camels, learn all about Ngaanyatjarra Camel Company's rules for air-camel transport. I admire their strong stance on not delivering camels to active war zones. It is good to have principles.
I learned about the Purdubik's Cube, a Purdue engineering school-designed robotic system that holds the Guinness World Record for fastest puzzle cube solve at 0.103 seconds (faster than a blink, requiring high-speed photography to capture).
Related: Last week, a 9-year-old Polish boy set the human record at 2.76 seconds. Purdue's robot could solve a Rubik's Cube 26.8 times in that span.
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.
Prompt:
A rodentologist in his 50s with gray hair, wearing practical work clothes (khaki pants, button-down shirt, work boots), crouches in a New York City subway tunnel. He holds a flashlight, its beam illuminating a brown Norway rat standing upright on its hind legs about three feet away. Both make direct eye contact in a frozen moment of genuine mutual curiosity and recognition.
Images:
Chat:
Gemini:
Perplexity:
Grok:
Epidemiologic Realities
This newsletter began during the pandemic. I leave these links here as a marker of 1) how to find resources on incidence and prevalence of various diseases and 2) to remind myself why I wear masks in my clinic and, often, on planes.
Lots of viruses in wastewater, and very high levels of coronavirus RNA in the upper Midwest.
Wastewater Scan offers a multi-organism wastewater dashboard with an excellent visual display of individual treatment plant-level data.
https://data.wastewaterscan.org/
https://data.wastewaterscan.org/
The Pandemic Mitigation Collaborative (PMC) uses wastewater viral RNA levels to forecast COVID-19 rates over the next 4 weeks.
https://pmc19.com/data/
https://pmc19.com/data/
Clean hands and sharp minds,
Adam
Comments
Post a Comment