iiinova — where your digital self keeps evolving

Brittany Trang @brittanytrang.com · 16/09/2026

I took a break from the newsletter this week, but so much has been happening that I asked my editor if I could send a very short edition today — he refused (which is probably for the best!) So here’s the roundup of STAT’s AI coverage from the past week: 🖥️🩺

110

Brittany Trang @brittanytrang.com · 20/07/2026

Seeing @ratboysband.bsky.social last night at Pritzker Pavilion on a 74°, sunny evening encapsulates everything that is good and perfect about peak summer Chicago. It makes me miss living here.

030

Brittany Trang @brittanytrang.com · 19/05/2026

If you haven't already tuned in to the #STATBreakthrough Summit, you've missed out! We started with a great interview with HW Bush & Clinton FDA commissioner David Kessler, who had messages for FDA staff, and the next FDA commissioner. Sign up for the stream here: www.statnews.com/stat-summit-...

010

Brittany Trang @brittanytrang.com · 03/05/2026

hello #chemsky #chemistry 🧪⚗️ friends; I have a question for you: Does ACS still give out the annual element mugs? I just got a letter asking me to renew for my 10th year and offering me Ne, but what I'm really after is a 9/F mug. I don't think I've gotten one since maybe 5 or 6. Anyone know?

343

Brittany Trang @brittanytrang.com · 08/04/2026

Coding intensity for patient visits is rising. Did AI scribes do that? You may have seen the Trilliant study that alleges AI scribes are making visit coding go up. But what does the study actually say? An analysis of that, and much more, in this week's AI Prognosis: www.statnews.com/2026/04/08/a...

123

Brittany Trang @brittanytrang.com · 08/04/2026

Happy 10th anniversary to Painting of a Panic Attack. RIP Scott.

Song of the Week: “Still Want to Be Here” by Frightened Rabbit
Ten years ago today, Scottish indie rock band Frightened Rabbit released what would become its final album.

No one was better at putting into words the complicated confluence of desire and dread than lead singer Scott Hutchison. That shines through on “Still Want to Be Here:” “I still want to be here, want to be here / And I would live in a devil’s ditch just to be near you.”

Hutchison died in May 2018, succumbing to the dark thoughts he often sang about struggling with.

040

Brittany Trang @brittanytrang.com · 02/04/2026

Please reshare! Two more #scicomm resources: "The beginner's guide to a career in science writing, 2.0," and a guide to applying to the AAAS Mass Media Fellowship. 🔗👇 Five years ago today (pictured), I found out that I got the AAAS MMF, which changed the course of my life.

Asian woman with glasses in a blue surgical mask takes a selfie, smiling. She is outdoors at golden hour on the sidewalk of a college campus.

2179

Brittany Trang @brittanytrang.com · 09/03/2026

Health insurers have claimed for months that AI is driving up health care costs. Now Blue Cross Blue Shield is offering proof. The ambiguous analysis methods would sow more doubt on the conclusion...if a hospital hadn't told me last year that its AI coding software was targeting this EXACT code. 🩺🖥️

2103

Brittany Trang @brittanytrang.com · 11/02/2026

It's AI Prognosis' birthday! 🎉 One year ago this week, I sent the first edition of @statnews.com's newsletter about AI in health care & medicine. I've learned a lot since then (and so have AIP readers!) Photo: Me, tired but proud, after launching the newsletter a year ago. 🖥️🩺

1113

Brittany Trang @brittanytrang.com · 06/02/2026

Proud of this contribution to today's Morning Rounds — and not just because I managed to get "squeezed from the teat of a beet" into the newsletter. #chemistry #food #fda

As STAT’s friendly neighborhood PhD chemist, I am here to remind you that everything is chemicals no matter where it comes from. “Natural” vanilla includes vanillin made from tree pulp waste, and just because a dye is “natural” doesn’t mean it’s necessarily squeezed from the teat of a beet. The FDA’s new working definition of “natural” dyes includes dyes that are manufactured via fermentation from yeast, algae, or fungi; mixed or reacted with other chemicals, or extracted from plants bred to maximize color production. —Brittany Trang

182

Brittany Trang @brittanytrang.com · 03/02/2026

You're tired of reading takes about Doctronic's Utah AI prescription pilot. But did you know that the company hasn't spoken to the FDA about it? Nine policy and legal experts STAT spoke to disputed Doctronic's argument that its AI is not a medical device: 🩺🖥️

“I’m befuddled why this is not a medical device and why folks believe that ‘medical device’ and ‘practice of medicine’ are mutually exclusive,” said Ateev Mehrotra, a physician and professor of health services, policy, and practice at Brown University School of Public Health.

130

Brittany Trang @brittanytrang.com · 03/02/2026

How HHS is implementing Trump AI mandates, in numbers I take a look at the AI tools the nation's health agency is using for everything from controversial audits to helping the agency cope with layoffs & executive orders. Full charts in the story @statnews.com: www.statnews.com/2026/02/03/n...

021

Brittany Trang @brittanytrang.com · 29/01/2026

1. Go to the general settings of your Gmail account 2. Leave "Smart features" ON but click on "Manage Workspace smart feature settings" 3. Toggle "Smart features in Google Workspace" OFF This does turn off "show events from Gmail in Calendar," so you may want to be cautious if you like that feature

141

Brittany Trang @brittanytrang.com · 29/01/2026

If anyone else's personal Gmail suddenly started giving AI summaries of your emails with no option to turn them off: I already took the elevated blood pressure hit for you and figure out how to get rid of it WITHOUT getting rid of the multiple inboxes (which I personally need in order to function)

2216

Brittany Trang @brittanytrang.com · 29/01/2026

Please reshare! Here's the first in a series of science writing resources I'm posting to my website: A list of science writing/communication internships & fellowships. brittanytrang.com/science-jour... #sciencejournalism #scicomm #scicommjobs #journojobs #sciencewriting #sciwri #stem

Screenshot reading:
Science Journalism Fellowships and Internships
This is a non-exhaustive list of science journalism / science communications internships and fellowships one might reasonably apply to with minimal formal experience (e.g., as a student) or with one fellowship (e.g., the AAAS Mass Media Fellowship) and some freelancing under one’s belt.

Science journalism fellowships
AAAS Mass Media Fellowship
Summer internship (June – August) for STEM undergrad students, grad students, postdocs, residents, deadline Jan 1
Based at sites around the country
Nature News
Two 6-month internships: Jan – June (deadline in October), July – Dec

11914

Brittany Trang @brittanytrang.com · 14/01/2026

Do you, like me, feel like the news is trying its best to kill you? Fear not, I am standing in front of the firehose of #healthAI 🩺🖥️ news for you. I present you with some tidy buckets of important health and biotech AI news you may have missed over the last month: www.statnews.com/2026/01/14/b...

Screenshot reading:
‘AI make number go up’ is going to be big this year

If you’ve heard me say it once, you’ve heard me say it a million times: AI in health care is going to make costs go up, not down.

Everybody is saying it now: In JAMA Health Forum, in NPJ Digital Medicine. HealthcareIT News reported that Intermountain got a seven-figure reimbursement boost in the first year of using Solventum’s AI clinical documentation integrity solution. Healthcare Finance reported that Baylor Scott and White Health expect to make an extra $42 million in net revenue after one year of using Accuity’s AI CDI and coding review tech, based on their data after six months.

and a sidebar for a story that says: Related Story

AI scribe companies promise to help doctors bill more. Who will pay the extra cost?

142

Brittany Trang @brittanytrang.com · 07/01/2026

This is how fast health AI is moving: Between when I drafted this post on Monday and when it sent this morning, two of these 2026 #health #AI predictions basically became outdated. (click in the below post to find out which ones they were!) 🖥️🩺

1154

Brittany Trang @brittanytrang.com · 31/12/2025

2. Seeing Hotline TNT play a laundromat. Will Anderson knows how to put on a show. He forced us to crowd into the tiny venue side of Dirty Dungarees, & he was right. He forced us to put away our phones after 3 songs, & he was right. It was loud. A pit opened up at the end of the show. 💯, no notes.

100

Brittany Trang @brittanytrang.com · 31/12/2025

My 10 favorite records from 2025, though I'm not wedded to this ordering. Instead of championing an album of the year, I'm trying a new thing: my 5 favorite music moments of the year 👇🧵

Screenshot reading:
2025 Top Albums
Raspberry Moon | Hotline TNT
Revelation | The Knocks/Dragonette
Losin' | Colin Miller
The Yips | Petey USA
Everything Must Go | Goose
Dog-Eared | Billie Marten
Patience, Moonbeam | Great Grandpa
Blood on the Silver Screen | SASAMI
Iron | Post Animal
Chain Yer Dragon | Goose

111

Brittany Trang @brittanytrang.com · 17/12/2025

Earlier this year, I asked more than 90 health AI experts what's on their bingo card for the rest of 2025. Who gets to say "I told you so" now? Our look back at what did and didn't happen in 2025: www.statnews.com/2025/12/17/2... 🖥️🩺

141

Brittany Trang @brittanytrang.com · 03/12/2025

We finally have randomized, controlled clinical trial results from ambient AI scribes. But what do these RCTs actually tell us? Look at the below chart. Time savings are often low, but doctors give the technology rave reviews, esp in anecdotes that people tend to believe more than data 👀 🧵 🖥️🩺

12520

Brittany Trang @brittanytrang.com · 26/11/2025

In July, I asked health #AI experts what was on their bingo cards for the rest of 2025. Now it's time to see who was right. Refresh yourself on the predictions: www.statnews.com/2025/07/02/w... Vote by December 10: www.statnews.com/ai-prognosis... 🖥️🩺

110

Brittany Trang @brittanytrang.com · 12/11/2025

Four years ago, we were @aaasmassmedia.bsky.social fellows, starry-eyed at our first full-time #sciencejournalism gigs. Now, we're working dream/beyond-dream jobs at the New York Times, @cenmag.bsky.social and @statnews.com. 2021 us could not have imagined 2025 us. #scicomm #stem #altchemjobs

192

Brittany Trang @brittanytrang.com · 08/11/2025

Thrilled to meet my longtime #scicomm hero @ehmee.bsky.social at #SciWri25 today and also shocked at how many people who would love her work have not heard of The Brain Scoop! you simply MUST watch her series documenting taxidermying a wolf!! (TW: gross) youtube.com/playlist?lis... 🧪

2150

Brittany Trang @brittanytrang.com · 05/11/2025

Amid multiple studies showing that AI models aren't using medical reasoning to score well on medical board exams, radiation oncologist/Harvard AI researcher Danielle Bitterman really nails how the thinking about LLMs is flawed: 🩺🖥️

“Models just being able to memorize an answer doesn’t mean that they can generalize that information to related and nearby tasks, which is basically what we assume when we test humans on these exams,” she said.

“We’re not assuming that they can only answer the questions on the exam — we’re saying, ‘Being able to answer the questions on this exam is a proxy for the general clinical knowledge base.’”

150

Brittany Trang @brittanytrang.com · 03/11/2025

That escalated quickly... The biggest story in health care AI right now has bypassed patient care and gone straight to the money. Insurers and providers are duking it out over bills and claims with AI lightsabers. I predicted this back in the summer 👇 Let's take a closer look. 🧵 🩺🖥️

Screenshot reading: But Robinson also warned the technology will escalate the ongoing war between providers and insurers on billing that’s already resulting in an AI arms race. As health insurance companies see their profit margins decline because members are getting more care than anticipated, “the payers are not going to simply pay you more now,” he said. “They can’t.”

1154

Brittany Trang @brittanytrang.com · 23/10/2025

Happy Mole Day to all the chemists out there (and others who care!) From 2021 me, who had access to a lab, here's exactly one mole of sodium hydroxide for your timeline: #chemsky #science #MoleDay 🧪⚗️

A lot of white pellets sitting on a piece of weigh paper on an analytical balance reading "40.0070 g"

180

Brittany Trang @brittanytrang.com · 20/10/2025

Thrilled to kick off NASW's #SciWri25 at the virtual plenary this Thursday! Our session will be about: - what AI actually is - what journalists need to know about it - ideas about how to use it in your reporting I'll be moderating the discussion at noon CT / 1 p ET on Oct 23. See you there!

ScienceWriters2025
Thursday Oct 23, 2025 / 12:00PM - 01:00PM

"The Truth About AI: How to report on it without giving in to hype, and how to use it in the newsroom"

073

Brittany Trang @brittanytrang.com · 17/09/2025

(Side note: these screenshots from a Pfizer ad the FDA complained about look like they were taken from a College Humor Hardly Working sketch....and when you look closer, that's literally Ross Bryant)

110

Brittany Trang @brittanytrang.com · 15/09/2025

Unfortunately using a "-ai" filter does not work for me, as I am a health AI reporter (subscribe to my newsletter!) and often need to search for AI stuff. Newsletter: www.statnews.com/signup/ai-pr... Also, adding "-ai" does not turn off the AI overview 🙃

Screenshot of Google search results page for the search "rate limiting step kinetics -ai" with an AI overview at the top

130

Brittany Trang @brittanytrang.com · 15/09/2025

Unfortunate news: I am here to tell you that the trick of including curse words in your Google search no longer disables the AI overview (I usually include a much milder word but had to experiment to see if the results were different...they were not)

Screenshot of Google search results page for the search "rate limiting step kinetics fuck" with an AI overview at the top

130

Brittany Trang @brittanytrang.com · 15/08/2025

Today is my 3-year STATiversary. I'll never forget getting an email from the NYT's Carl Zimmer as I took the Red Line to the STAT office the first time (pictured below!), asking me for an interview about my Science paper. Weird way to start off a career as a reporter—getting interviewed yourself! 🧵

2211

Brittany Trang @brittanytrang.com · 23/07/2025

The most intriguing idea is the proposal to explore sequencing all life on federal lands to create data for a biological foundation model (sounds kind of like Basecamp Research's model) 🩺🖥️🧬🧪

Explore the creation of a whole-genome sequencing program for life on Federal lands, led by the NSTC and including members of the U.S. Department of Agriculture, DOE, NIH, NSF, the Department of Interior, and Cooperative Ecosystem Studies Units to collaborate on the development of an initiative to establish a whole genome sequencing program for life on Federal lands (to include all biological domains). This new data would be a valuable resource in training future biological foundation models.

110

Brittany Trang @brittanytrang.com · 23/07/2025

This is also an interesting (maybe controversial) idea: Emphasizing datasets that are amenable to AI by using "what's the impact of datasets from your previous work" as a metric for grant funding, and requiring people to disclose datasets that are used in AI work that's federally funded

• Incentivize researchers to release more high-quality datasets publicly by considering the impact of scientific and engineering datasets from a researchers’ prior funded efforts in the review of proposals for new projects. • Require federally funded researchers to disclose non-proprietary, non-sensitive datasets that are used by AI models during the course of research and experimentation.

100

Brittany Trang @brittanytrang.com · 23/07/2025

There are some interesting #science proposals in the White House's new AI Action Plan (pgs 11-12). A thread: 🧵 First, the administration wants to invest in the "cloud labs" concept: 🖥️🩺🧪⚗️🧬⚛️🔬 #sciencesky #chemsky www.whitehouse.gov/wp-content/u...

Invest in AI-Enabled Science Like many other domains, science itself will be transformed by AI. AI systems can already generate models of protein structures, novel materials, and much else. Increasingly powerful general-purpose models show promise in formulating hypotheses and designing experiments. These nascent capabilities promise to accelerate scientific advancement. They will only do so, however, with critical changes in the way science is conducted, including the enabling scientific infrastructure. AI-enabled predictions are of little use if scientists cannot also increase the scale of experimentation. Basic science today is often a labor-intensive process; the AI era will require more scientific and engineering research to transform theories into industrial-scale enterprises. This, in turn, will necessitate new infrastructure and support of new kinds of scientific organizations.

210

Brittany Trang @brittanytrang.com · 23/07/2025

In this week's AI Prognosis, one VC's take on what health AI areas are good/bad investments right now Plus: I'm confused about CHAI's Medicaid AI "tiger team." Are you? Reach out: brittany.trang@statnews.com / btrang.01 on Signal www.statnews.com/2025/07/23/w...

Emma Beavins at Fierce Healthcare has more details on the proposed workgroup, which will be working on a compressed timeline: Six months for a draft, and finalized within a year, in an effort to keep pace with HHS’s implementation deadline. That means that it will “likely” skip some steps that are usually included in the CHAI process, like community review and public comment.

“This is a little bit more of a nuanced area where we may not need as much rigor as one of our more formal best practices,” CHAI CEO Brian Anderson told Beavins.

How can something that is nuanced not need as much rigor as something that is less nuanced? When STAT asked CHAI for clarification, Anderson said that by “nuanced,” he was referring to the “emerging nature of” AI uses for Medicaid. “We will continue to advocate for oversight given the equity implications, but we’re also building guidance in a space where there just isn’t a deep bench of prior work to pull from,” he wrote in an emailed statement.

But doesn’t the relative lack of information mean that guidelines in this space should require more study, not less? Are you also confused about this initiative? I would love to hear from you (CHAI members and non-members alike). I’m open to off-the-record or on-background conversations: brittany.trang@statnews.com, or btrang.01 on Signal.

120

Brittany Trang @brittanytrang.com · 23/07/2025

The news about Wired changing up its subscription reminded me: This weekend I became a @wired.com subscriber, and also became a monthly donor to @npr.org and my childhood @pbs.org station. If you want something to keep existing, now is the time to support it!

070

Brittany Trang @brittanytrang.com · 17/07/2025

Also, for any of the @dimension20.bsky.social #cloho #cloudwardho fans out there... I kept thinking about the walking snake the entire time I was writing this, but that analogy would have been even more complicated than pasta

010

Brittany Trang @brittanytrang.com · 09/07/2025

The mysterious "American Society for AI" gives me the same vibes as the Yale secret society from Gilmore Girls. It's a "private & prestigious club that brings together prominent leaders & executives" to ‍"create a better world with AI." 🤨 Check out its rules:

FAQs:
• Only members can attend the gatherings
• Membership is by invitation only
• We do not share membership information
• Members cannot disclose the names of other members
• The gathering location is provided only to members
• No recordings are permitted during the gatherings
• We decline any sponsorships and donations
• We decline any requests from media/press

If you have a question, please contact a board member.

110

Brittany Trang @brittanytrang.com · 02/07/2025

What's on your health AI bingo card for the rest of 2025? I asked more than 90 health and life sciences AI experts for their 🌶️ takes as we kick off the second half of the year at AI Prognosis: 🖥️🩺 #health #medicine #AI www.statnews.com/2025/07/02/w...

111

Brittany Trang @brittanytrang.com · 27/06/2025

The classic "brag about reducing prior auth on Monday" / "introduce new prior auth program on Friday" 🤘 Dr. Oz and CMS introduce a new, tech-enabled Medicare insurance approval program for fraud and waste, just days after holding a press conference talking about reducing prior auth.

162

Brittany Trang @brittanytrang.com · 18/06/2025

Lots of detail in Lizzy's piece about how Prasad is going outside of the norms for the the CBER role already and acting more like a political appointee. The role is traditionally for career scientists who rise through the ranks and oversee a fair amount of day-to-day operations. 🩺💉💊

1146

Brittany Trang @brittanytrang.com · 28/05/2025

If this doesn't entice you to click, today's edition of AI Prognosis also includes: @andrewedunn.bsky.social's analysis of AI vs CRISPR hype and Future House's latest exploits a recommendation for a MJ Lenderman-adjacent album of the year contender a recent study that prompted this meme:

Meme of Anna Garcia on Dropout's show Make Some Noise. Anna is labeled as "People using AI for medical advice" and is covering up, with her hands, part of a Rorschach blob labeled "AI's ability to confidently say things that aren't true." The still is captioned "I'm sort of ignoring this."

050

Brittany Trang @brittanytrang.com · 23/05/2025

And asking ChatGPT (no RDKit in the prompt) to interpret the SMILES strings by itself was the worst. It's not even close. It also was unable to generate a non-cropped image, even after it agreed to fix that, so this is what we've got:

170

Brittany Trang @brittanytrang.com · 23/05/2025

Then I realized that maybe I should be using SMILES strings to prompt it, as Brockman did, even though I've never seen a chemist do that. Because I ran out of free ChatGPT entries, @amanda-garces.bsky.social ran some for me with GPT-4o. Using SMILES and RDKit, close but no cigar:

160

Brittany Trang @brittanytrang.com · 23/05/2025

I realized I didn't ask it to use RDKit to draw the molecules, as OpenAI president Greg Brockman did in his example in his post on X. So I tried that. It gave me this, which is actually correct as far as I can tell! But the problems are obvious, even to someone who doesn't know chemistry:

291

Brittany Trang @brittanytrang.com · 23/05/2025

Here's what ChatGPT gave me. Besides these molecules not being right and it not following Buchwald-Hartwig logic, these structures are just not how you draw molecules 🤦‍♀️ Like, what's up with that CH just hanging out there? #chemsky #AI 🖥️🧪⚗️

2182

Brittany Trang @brittanytrang.com · 23/05/2025

But when I asked it "can you draw the structures of the starting material and product for the coupling of 2-bromo-5-formyl thiophene and tert-butyl 4-aminobenzoate," it absolutely could not execute. Here are the correct structures, drawn by me:

161

Brittany Trang @brittanytrang.com · 23/05/2025

As with other AI models I've tried, I asked it to give me conditions for coupling 2-bromo-5-formyl thiophene & tert-butyl 4-aminobenzoate. It did a really good job of giving me a text procedure for a Buchwald-Hartwig. It even identified that the aldehyde is sensitive (I learned this the hard way)

161

Brittany Trang @brittanytrang.com · 23/05/2025

OpenAI just updated ChatGPT to be able to use RDKit, a cheminformatics Python package. OpenAI's president says this makes ChatGPT "useful for scientific work across health, biology, and chemistry," but it is hilariously still not good at chemistry (🧵) #chemsky #AI ⚗️🧪🖥️

88026