You’re reading the web version of STAT’s Health Tech newsletter. This newsletter is a guide to how technology is transforming the life sciences. Sign up to get it delivered to your inbox every Tuesday and Thursday.
President Trump rescinds order on artificial intelligence
Since President Trump took office, our attention has focused on the sweeping changes pushed through by executive orders, and as expected, President Trump rescinded former President Biden’s order on artificial intelligence starting in 2023. At the time, STAT’s Casey Ross summed up the meaning of Biden’s presidential election. For health AI developers. The most well-known provision requires developers of basic models that “may pose a risk to national security” to provide information to the federal government, including information about safety testing. This was incredibly unpopular in the industry.
The order also made a number of demands on the Department of Health and Human Services, including developing a just-submitted strategic plan for artificial intelligence and creating an inventory of ways the department could leverage AI. . Despite the recent excitement surrounding AI, the federal government and Congress have been slow to do much about guardrails around the technology in health care and other fields, although regulators have put in place some limited rules. I didn’t see any progress. It’s unclear what exactly the Trump administration will do, but it will certainly take a more interventionist approach.
Marketing Overview of Formulated GLP-1
GLP-1 is the hottest drug of this generation, and if you’re a discerning person, you’ve probably been bombarded with marketing about the huge benefits of the popular obesity and diabetes drug. But are you sure that everything you read online is true?
Drugs are expensive, demand has led to shortages, and online services are rushing to fill the gap with low-cost compounding drugs. As STAT’s Katie Palmer reports, a new study found that of 79 websites selling compounded GLP-1 or its prescriptions, 37% said the drug was FDA-approved. It turns out that it states or implies that there is. Nearly half did not include information about the drug’s side effects, warnings, or contraindications. Composite copies are not approved by the FDA.
The study highlights concerns about the proliferation of compounded GLP-1 as shortages ease and drug companies consider ways to offer branded drugs at lower prices.
Click here for details
model card madness
Under a federal deadline at the end of the year, developers of certain AI products used in healthcare must provide end users with details about how they are developed and used. The use of “model cards” or “nutrition labels” detailing machine learning and other technologies is widespread in the broader technology industry, but its use in healthcare is new. So to help developers comply with the new rules, or simply to help spread the word about their products, the Coalition for Health AI and the Duke Institute for Health Innovation are each publishing templates that are free to use. Masu.
Interestingly, the Food and Drug Administration does not actually require developers of regulated AI medical devices to use model cards to describe their products in detail, although the AI developer announced earlier this month The draft guidance document for 2017 states that model cards may be useful and includes examples. Among the industry’s concerns is that conflicting regulatory regimes make compliance difficult, so it will be interesting to see what efforts are made to harmonize recommendations across federal health agencies. It’s going to be interesting.
WuXi AppTec unloads medical device test unit
Brittany Trang writes: Last week, China-based Wuxi AppTec sold its US medical device testing business to private equity-owned US company NAMSA. The move is the latest in a series of Wuxi sales after the BIOSECURE Act passed the U.S. House of Representatives last fall. The law would prohibit pharmaceutical and biotech companies from using equipment and services from foreign companies of “concern” (understood primarily as Chinese companies such as Wuxi Uptech and Wuxi Biologics). It was. However, the bill has been left out of the year-end policy and is likely to be more difficult to pass under the new administration. As part of the deal, NAMSA will take over WuXi’s locations in Minnesota and Georgia.
EevolutionaryScale’s new protein model
Brittany also said: EevolutionaryScale, a protein AI company founded by former Meta protein AI employees, unveiled a new protein language model last week that can handle not only a protein’s sequence, but also its structure and function. The model, ESM3, was previously published as a preprint co-authored by Arc Institute researchers last summer. To show that the model can generate sequences that differ from known ones while maintaining functionality, the researchers used it to create a version of the green fluorescent protein that was only 58% similar to other versions. Generated. EevolutionaryScale aims to bring protein design AI technology to other companies and has raised $142 million in a seed round in 2024 with participation from Amazon and NVentures, the venture capital arm of chipmaker NVIDIA. .
Want more health AI news? Sign up for Brittany’s AI Prognosis newsletter in February.
what we are reading
Amid looming crisis in infant formula market, STAT online therapy boom is primarily benefiting the privileged, study finds, New York Times Sutter Health develops ambitious expansion plans for AI imaging technology Strikes deal with GE Healthcare, fierce healthcare President Trump orders US to withdraw from World Health Organization STAT