Some nurses objected to AI that determines nurse staffing based on real-time charting. Another nurse complained about an AI chatbot that patients can talk to instead of a nurse. 'What if a patient is having a heart attack & the chatbot doesn't understand?' beckershospitalreview.com/healthcare-inf…

Wow: 'Nurses are all for tech that enhances our skills and the patient care experience. But what we are witnessing in our hospitals is the degradation and devaluation of our nursing practice through the use of these untested technologies.' beckershospitalreview.com/healthcare-inf… via @beckershr

🦙 Exciting news in open AI (not OpenAI) - Meta just introduced Llama 3, their next-gen open source large language model. A few interesting notes:
1. Fun to read that the model team didn't even have access to their own high-quality human eval set used to assess real-world

The 'State of AI in 13 Charts' is a great way to get a quick summary of this year's AI Index. hai.stanford.edu/news/ai-index-…

#AIIndex2024 Stanford HAI Stanford AIMI

This entire report is a fantastic resource to understand recent progress in AI, but be sure to check out the new chapter on medicine and scientific discovery. aiindex.stanford.edu/report/?sf1877… Stanford AIMI

🎉 Celebrating 6 years at Stanford AIMI! We're grateful to our partners & supporters, near & far, dedicated to advancing health through AI for everyone. Thank you for being an integral part of our community. Here's to more breakthroughs & continued innovation! #StanfordAIMI

Delighted to be speaking to Stanford Med Alumni Day this morning about how AI will change medicine. Shout out to the class of ‘89—can’t believe it’s been 35 years!

Stanford AIMI

As part of Stanford Medicine's 'Transforming Healthcare Through Big Data, Analytics & #AI ,' we met with learners from around the world to share AIMI faculty & industry affiliate insights into tech's impact on healthcare leadership, management & transformation! stanford.io/3TM2wfq

Some interesting use cases for #GenerativeAI
Most of them are for helping people and making them more productive.

I’m excited to finally showcase our work on developing a generalizable deep learning system for cardiac MRI. This is the culmination of over 3 years of research spanning my time as a postdoc in Hiesinger Lab and now as a cardiac surgery resident at PennSurgery, and I couldn’t be

This is a fantastic idea that bridges a major gap. We need the same for the many non-EMR-based AI algorithms. Because: “a critical access hospital in rural Nebraska sees a different mix of patients and has different workflows than a dedicated cancer center in New York City”

'To provide greater transparency regarding how U.S. FDA medical product centers are collaborating to safeguard public health while fostering innovation, CBER, CDER, CDRH, OCP are jointly publishing this paper.'
fda.gov/media/177030/d… (PDF)

@stanfordAIMI Stanford HAI

inspired by this famous study (making fun of MRI researchers) psychology.mcmaster.ca/bennett/psy710…

If you feed AI an MRI, it will happily write a detailed and very convincing diagnosis...

even if the patient is a dead salmon.

That is awesome—nice work! I was trained to read CXRs by radiologists who learned before cross-sectional imaging existed. They taught us how to recognize enlargement of each chamber—not taught as much today. AI carries on the long tradition, with even more well-honed skills.

Who benefits from AI?

Our new study with 140 radiologists in Nature Medicine challenges prevailing assumptions.

We find years of experience, subspecialty and familiarity with AI tools fail to reliably predict the impact of AI assistance.

nature.com/articles/s4159…

And we're off! #RadAdv has posted its first articles. Continuous publication minimizes delays. RSNA Oxford Academic MGH Radiology Chiefs

academic.oup.com/radadv

A comprehensive assessment of privacy preserving ML in medicine:
'Privacy-preserving training yields lower accuracy, but largely does not discriminate against age, sex or co-morbidity. Difficult diagnoses & subgroups may suffer stronger performance hits'
nature.com/articles/s4385…

Given the severity of the workforce shortage, no single approach will fix it. We need an “all of the above” strategy: AI, imaging order-entry decision support, advanced practice providers working under the supervision of radiologists, etc.

The issue is increasing imaging volumes, older rads retiring, while residency slots don’t keep up. As new diagnostic and therapeutic imaging technologies continue to become available, imaging becomes useful in new settings, often replacing more invasive alternatives.