It's Dr. Oz's turn to tout health AI on the Hill

You’re studying the net version of STAT’s Well being Tech publication, our information to how know-how is reworking the life sciences. Signal as much as get it delivered in your inbox each Tuesday and Thursday.

On Friday, Mehmet Oz, President Donald Trump’s choose to run the Facilities for Medicare and Medicaid, turned on the charisma and cruised by way of a affirmation listening to earlier than the Senate Finance Committee that featured loads of substantive dialogue of coverage. For a wonderful evaluation learn STAT’s Tara Bannow’s protection of the listening to. Well being tech, together with telehealth and AI, got here up just a few occasions:

On AI, Oz made loads of statements most individuals would discover completely uncontroversial, together with in regards to the potential of know-how to reinforce care and cut back paperwork for clinicians, and in regards to the want for guardrails to make sure dangerous issues don’t occur.

Oz famous that AI might be used to hurry up pre-authorization for medical procedures. “We have now AI assist instruments, navigation techniques that might fairly shortly adjudicate whether or not it is best to have to attend even a day to get the treatment that can get you out of ache, or perhaps a week for the process that you have to be allowed to have,” he mentioned, “It’ll take loads of the angst out of the system for the American folks.”

Pressed about AI being utilized by Medicare Benefit insurers to make determinations about whether or not to disclaim care, Oz urged AI might be used to identify the issue: “We needs to be utilizing AI inside the company to establish that early sufficient in order that we will forestall it.”

Evaluating AI on well being duties that really matter

Final yr, a overview of 519 well being care massive language mannequin research revealed between 2022 and 2024 confirmed that 45% of them examined the synthetic intelligence fashions’ information by asking them multiple-choice questions like these you may see on the U.S. Medical Licensing Examination. However performing effectively on multiple-choice questions doesn’t essentially translate to the duties LLMs may be requested to finish in the true world, like with the ability to summarize a affected person document, give you acceptable billing codes, or diagnose a affected person.

In a brand new story, Brittany Trang seems to be at a brand new AI language mannequin analysis framework from Stanford and talks to Stanford Well being Care chief information scientist Nigam Shah about how the brand new analysis framework — which the crew has already run on fashions like Google’s Gemini 1.5, OpenAI’s GPT-4o, and Meta’s open-source Llama — will help well being techniques determine which fashions to make use of for the well being care duties that they care about.

Learn extra right here

Why disabled individuals are in danger from AI

O. Rose Broderick writes: A brand new report highlights how well being applied sciences that depend on AI and algorithmic techniques can fall quick for folks with disabilities. The report is one other pink flag about how a sufferers’ identities can have an effect on the care they obtain from well being applied sciences and the way such techniques typically battle to adequately serve folks from marginalized communities. Till not too long ago, a lot of the analysis on this area has centered round race and gender, however extra researchers are turning their eyes in direction of points that have an effect on practically 20% of the inhabitants — Individuals with disabilities.

“For folks with disabilities, particularly, there’s an actual threat of discrimination. Applied sciences are usually not being developed, deployed, and typically audited in methods which are inclusive of individuals with disabilities,” mentioned Ariana Aboulafia, the incapacity rights in know-how coverage lead on the Middle for Democracy and Know-how.

Learn the entire interview right here

Google’s massive well being occasion preview

Right this moment, I’m heading to Google’s annual Test Up occasion in New York, the place the corporate will share he newest about how its applied sciences, together with client merchandise and enormous language fashions are being utilized to well being. I’ll have extra to say within the Thursday version, after the occasion, however Google this morning revealed a preview of what the corporate shall be discussing. You may stream the occasion your self at 11 a.m ET.

In search, Google wrote that it has made enhancements to its AI Overviews for well being associated matters in order that they’re “extra related, complete, and proceed to fulfill a excessive bar for scientific factuality” because of “current health-focused developments on Gemini fashions.” At Google and elsewhere there’s an growing concentrate on ensuring that generative AI merchandise present dependable solutions to health-related questions.

Google additionally introduced a brand new replace that might make it simpler for customers of Android telephones to entry details about care they acquired from suppliers. The corporate’s Well being Join app, that enables customers to sync completely different varieties of information between apps, has a brand new API that enables apps “to learn and write medical document info like allergy symptoms, medicines, immunizations and lab ends in normal FHIR format.”

Google will discuss in regards to the lack of pulse detection function for Pixel smartwatches and its new AI co-scientist, each of which we’ve mentioned not too long ago in STAT newsletters. Info distributed upfront makes no point out of well being fairness, as anticipated, after the corporate not too long ago dumped the idea in favor of “well being optimization” as President Trump assaults range, fairness, and inclusion each within the federal authorities and throughout the personal sector.

In an announcement, little question timed to the occasion, Quest Diagnostics revealed a brand new cope with Google Cloud that features utilizing information analytics and generative AI applied sciences to “give sufferers personalised well being insights.”