I've been thinking about the "virtual cell" concept and wanted to write up a few thoughts. Specifically on how I think the prior experience in GWAS informs the most likely way these models will be useful. andrewcarroll.github.io/2025/12/23/t...

Definitely true in medicine: When the costs of failure are high, the demo to product interval is long (e.g., Waymo).--Andrej Karpathy on the @dwarkesh.skystack.xyz podcast.

📢 This just in: Round 2 of "The Digitalist Papers," a deep dive into the economics of #AI from @erikbryn.bsky.social's @stanforddel.bsky.social. Essays from SIEPR's Susan Athey and Ramin Toloui, among others. Link below.

Catch JCHI Exec Director @kdpsingh.bsky.social @ucsdhealth.bsky.social Dec 10, 1 p.m. PST for a virtual keynote at Stanford AI+HEALTH Conference. A great lineup on the latest in AI and healthcare. Register ⬇️ medhub.stanford.edu/event/aiheal...

Thanks @rasushrestha.bsky.social—great way to find our radiology peeps on @bsky.app! #RSNA25

🚨 Agentic Systems in Radiology Everyone’s talking about agents🕵️‍♀️🕵️‍♂️ — but what happens when you bring them into real clinical workflows? Our new preprint explores the design, applications, challenges, and evaluation of LLM-based agents and agentic workflows in radiology 👇 arxiv.org/abs/2510.09404

Yes! A model trained with high quality healthcare data is usually better at a specific healthcare task than a general language model adapted to that task. www.nytimes.com/2025/10/16/o...

Don’t miss the free virtual MIDRC Seminar on Oct. 21, 2–2:45 p.m. CT! This month’s topic, Foundation Models in Radiology, will be presented by Drs. Curt Langlotz & Akshay Chaudhari. https://bit.ly/3Vw5GUz

Radiologists are seeing significant reductions in radiograph reading times with the use of artificial intelligence-generated reports 📖 #radiology @rasushrestha.bsky.social @jimrawsonmd.bsky.social @woojinkim.com radiologybusiness.com/topics/artif...

👇

In this Stanford news article, @stanfordhai.bsky.social Senior Fellow @suryaganguli.bsky.social argues that universities are essential for understanding how AI works. Here, he outlines three reasons why: news.stanford.edu/stories/2025...

📸 Today from DC: @stanfordhai.bsky.social Faculty Affiliate Michelle Mello testified in Congress on AI in healthcare. To boost AI adoption, she discussed policy changes that could build trust in AI's performance. Read her testimony here: bit.ly/4ngI56k

Good to see @bsky.app has emerged as the social media platform of choice for science stuff arstechnica.com/science/2025...

The phrase "artificial Intelligence" was born exactly 70 years ago. Wow. 🤯

Some generative A.I. tips for patients www.wsj.com/opinion/a-ch...

📣 Announcing the AI for Organizations Grand Challenge, a new competition for scholars to help organizations enter the era of AI. GoogleDeep Mind and @stanfordhai.bsky.social invite researchers from any university worldwide to submit your boldest ideas. Learn more: hai.stanford.edu/aiogc

Congratulations to HAI Senior Fellow @suryaganguli.bsky.social on leading the new @simonsfoundation.org Collaboration on the Physics of Learning and Neural Computation! The collaboration will study the fundamental scientific principles underlying AI. www.simonsfoundation.org/2025/08/18/s...

RadGPT helps patients understand their radiology reports. “We hope that our technology won’t just help to explain the results, but will also help to improve the communication between doctor and patient,” said @curtlanglotz.bsky.social, senior author of the study. hai.stanford.edu/news/new-lar...

Medicine needs Neuro-Bayesian AI (paywall): www.wsj.com/articles/mee...

Good timing too as the FDA yesterday released its updated list of AI-enabled medical devices that have received marketing authorization.

Such important work! www.nature.com/articles/s41...

📣 Our new PadChest-GR benchmark in collaboration with @ua.es and MedBravo is published in @ai.nejm.org: ai.nejm.org/doi/full/10....! 📰 Also check out our blog post for why we're so excited about it: www.microsoft.com/en-us/resear... 💾 The dataset can be downloaded here: bimcv.cipf.es/bimcv-projec...

Oops

A new large language model called “RadGPT” can help patients understand their radiology reports. Learn how a team of Stanford researchers developed the model: hai.stanford.edu/news/new-lar...

We're taking a big step towards medical superintelligence. AI models have aced multiple choice medical exams – but real patients don’t come with ABC answer options. Now MAI-DxO can solve some of the world’s toughest open-ended cases with higher accuracy and lower costs.

This research is just the first step on a long, exciting journey. We’re excited to keep testing and learning with our healthcare partners in pursuit of better, more accessible care for people everywhere. More on the blog today: microsoft.ai/new/the-path...

The National Academy of Medicine Shared Commitments provide common-ground expectations for the integrative benefits of learning health systems, which align science, incentives, and culture to facilitate continuous improvement. Full Sounding Board article: nej.md/44js5Zv @nam.edu #MedSky

The promise and problems with multi-cancer early detection tests by Sid Mukherjee @newyorker.com www.newyorker.com/magazine/202...

FDA issues final order reclassifying radiological CAD software www.auntminnie.com/imaging-info...

Congratulations to @erichorvitz.bsky.social and team—such nice work. AI + human is still better than either one alone if the collaboration is well designed.

Congratulations to the authors—such nice work. AI + human is still better that either one alone if the collaboration is well designed.

Paper, soon to appear at #ACL2025 main: arxiv.org/pdf/2505.24223 Project page, with all resources (datasets, models, ontology) and usage notes: stanford-aimi.github.io/srrg.html All models and datasets are publicly available as open-source: huggingface.co/collections...

💥 We unveil our paper accepted at the #ACL2025 Main Conference: Automated Structured Report Generation Let's revisit automated radiology report generation for CXR. Free-form reports make it hard for AI systems to learn accurate generation, and even harder to evaluate. 🧵👇 @StanfordAIMI @hopprai

In 2018, I had the immense pleasure of participating in #RSNAFast5! Now, I have the honor of serving as the moderator at #RSNA25! Apply to be one of 5 selected to deliver a five-minute talk in the Arie Crown Theater on a non-clinical topic. To apply now 👉 bit.ly/4kAxDFa @rsnasky.bsky.social

"People using an algorithm that offered recommendations only in cases where a human was likely to be uncertain or incorrect made the most accurate decisions, outperforming those using a predictive algorithm as well as those using no algorithmic support." news.stanford.edu/stories/2025...

Welcome!!

Submissions are now open for #RSNAFast5! Share your insights with the #radiology community by submitting your non-clinical topic today! You may be one of five selected to deliver a five-minute talk in the Arie Crown Theater at #RSNA25. Submit by June 23.. bit.ly/4kAxDFa

I left last year November leaving behind a huge following I have much less here but I’m ok with that and in a much better place

Pivots also can indicate that the primary business isn’t particularly healthy, which wouldn’t be surprising given the suite of medical imaging AI products currently on the market.

The papers all follow a similar template and make misleading claims https://go.nature.com/4kucwEH

@cekahn.bsky.social, editor of Radiology: Artificial Intelligence, weighs in on how radiologists are harnessing the power of AI—while recognizing its current limitations. His insights highlight the critical role of expert interpretation in the age of AI. www.nytimes.com/2025/05/14/t...

"This notice alerts the extramural community that NIH has temporarily disabled the No-Cost Extension functionality in eRA Commons...At this time, all requests for NCEs must be submitted as a prior approval request in eRA Commons, for NIH review and approval." grants.nih.gov/grants/guide...

!!

Do foundation models in medical imaging see everyone fairly?🤔 Excited to share our new Science Advances paper uncovering & auditing demographic biases of expert-level VLMs, and comparing to board-certified radiologists🧑‍⚕️ 📄science.org/doi/10.1126/sciadv.adq0305 💻github.com/YyzHarry/vlm-fairness (1/)

RadLex, a comprehensive radiology lexicon developed by RSNA, celebrates its 20th anniversary this year. bit.ly/4lGFHFK

Nobody should be surprised: Sociodemographic biases in medical decision making by large language models www.nature.com/articles/s41...

AI hasn’t killed radiology, but it is changing it "if I’m the average radiologist — and most people are average — when you use AI, you become an expert. Who benefits from that? The patients.” www.washingtonpost.com/health/2025/... #medsky #radsky @curtlanglotz.bsky.social @rsnasky.bsky.social

Think you’ve got your finger on the pulse of AI? The #AIIndex2025 is out—and it comes with a quiz! Test your knowledge and share your results: hai.stanford.edu/news/how-wel...

📢Introducing #AIIndex2025: This year's report highlights the most critical trends in AI – from shifting geopolitical landscape and rapid technological evolution, to AI’s expanding role in science and medicine, business, and public life. Read more: hai.stanford.edu/ai-index/202...

wapo.st/4lkJoAV @ctisus.bsky.social @rajpurkar.bsky.social Despina Kontos, Christoph Wald