Will patients drive demand for AI in health care?

It's been a whirlwind of health and AI conferences on the West Coast this past week, and I've heard industry veterans, bioethicists, software developers and patients debating recurring foundational questions: Should we allow AI to actually diagnose patients? Do patients even want the technology to encroach on medical visits? And who's liable — the doctor or the software developer — if AI leads clinicians astray and harms patients?

Fortune Brainstorm Health is overflowing with industry luminaries like Included Health's cofounder and CEO Owen Tripp, Centene's CEO Sarah London, and Advancing Health Equity's CEO Uché Blackstock; onstage, experts tackled some of these questions and others. A few tidbits from the panels:

Asked for a single solution to the country's chronic disease crisis, Omada CEO Sean Duffy suggested easing payment for asynchronous care, a service his company offers for conditions like diabetes.
Noosheen Hashemi, who founded and heads January AI, offered up AI-enhanced behavioral treatment and psychedelics as potential ingredients of early preventive care, prompting murmurs in the audience. (What do you think?)
Centene's London said the company had temporarily been able to reach some patients about Medicaid redetermination thanks to a Federal Communications Commission waiver, but called for a program-wide modernization allowing managed care organizations to reach Medicaid patients where they are.
On AI for diagnoses, Andreessen Horowitz general partner Julie Yoo posited that ambient medical documentation tech that automatically generates medical notes are essentially already "creating a proposal for diagnosis."
Isaac Kohane, a Harvard biomedical informaticist and editor-in-chief of the journal NEJM AI, predicted patients would soon demand AI in their medical care.

A16z's Yoo talks noisy AI regulations

I snagged Yoo before her panel for a quick temperature-check on generative AI and Washington's attempts to rein in its potential harms. "There's a lot of noise," she said, warning policymakers against a "regulatory sandwich" that might introduce confusion among startup founders about "which of the 10 regulations that have been laid out in front of me actually pertain to me." Andreessen Horowitz, she said, is squarely on the side of "little tech": smaller, scrappy ventures that may not have the resources to keep up with regulatory creep.

(...Also on AI regulation, the Coalition for Health AI's chief executive Brian Anderson published an op-ed in Nature Medicine defending the role of quality assurance labs in validating AI, while also emphasizing that such a model could "serve the whole population, not just the wealthy." Critics have argued that concentrating testing centers mainly at large academic medical centers could deprive lesser-resourced and rural organizations of validation tools.)

ICYMI: #STATBreakthrough Summit West

Late last week STAT convened its annual West Coast summit, where health industry giants appeared on-stage and off. (23andMe CEO Anne Wojcicki was among audience members, and shared her thoughts on generative AI here.) Other highlights you may have missed:

STAT_Breakthrough_SF_6-768x432-jpg

Peter Lee (pictured above) who heads Microsoft Research, argued that ChatGPT and similar tools shouldn't be used for initial diagnoses — at least until bioethicists agree on appropriate use and safeguards. He also described the dread he felt when a doctor friend trotted out an earlier, less sophisticated version of a generative AI application to write up patient notes. (More on his conversation with my colleague Nicholas St. Fleur from me here. And a write-up on Stanford RAISE Health's symposium, where I talked to Lee, Verily co-founder Jessica Mega and Stanford's Sylvia Plevritis earlier that week, here.)
I asked Venrock partner Bob Kocher, who routinely gets unwelcome pitches on public transit and at the gym, about the worst AI ideas he's fielded. They're often related to dating, providing users with a reliable, on-demand source of conversation, he said. Read more on our chat, which ranged from AI regulations to Change Healthcare and General Catalyst's bold bet on a health system, from Lizzy. (Spoiler: He thinks General Catalyst's big swing is, at best, misguided.)
Abridge CEO and co-founder Shiv Rao demonstrated his startup's rapid note-taking capabilities; asked by an audience member how AI might exacerbate or mitigate racial bias frequently reflected in clinicians' notes, MemorialCare Health System's David Kim pointed out AI could eventually help organizations could gather detailed analytics on notes shown to contain obvious bias, for instance.
In conversation with Mario Aguilar, UCSF's Ida Sims said wearables generate powerful data but that such data isn't all that useful to clinicians — yet. (Read more from Brittany Trang here.)

Washington

Second House panel advances two-year telehealth extension

Back in Washington, late last week a House subcommittee voted to advance a measure to extend some Medicare flexibilities covering virtual care for two-years — and that puts lawmakers in alignment with another House committee, reports Rachel Cohrs. The House Energy & Commerce's health panel had previously examined a bill that would permanently extend those flexibilities, but on Thursday amended the bill to continue the expanded coverage for just two years. The House Ways & Means Committee has already passed a two-year extension. Read more on the legislative prospects here.

industry news

Hims climbs on the the GLP-1 bandwagon

A flurry of health tech product announcements dropped this week: Direct-to-consumer telehealth pioneer Hims & Hers Health is now offering GLP-1 injections for $199 a month, which includes unlimited visits with a provider. Also of note:

Uber Health has a new feature allowing patients' caregivers to coordinate medical transport and grocery delivery, and also to help patients pay using health benefit cards.
Cityblock is partnering with Florida managed care plan Sunshine Health — a subsidiary of Centene — to offer primary care and care coordination to Medicaid patients in several central Florida counties. It's the startup's first partnership in Florida, but the third state it's operating in under its broader partnership with Centene in addition to New York and Ohio.

More around STAT

Check out more exclusive coverage with a STAT+ subscription

Read premium in-depth biotech, pharma, policy, and life science coverage and analysis with all of our STAT+ articles.

What we're reading

OpenAI's longterm AI risk team has disbanded, WIRED
Black women founders offer health tech solutions to combat maternal mortality, STAT
Oracle's deadly gamble, Business Insider

Thanks for reading! More on Thursday - Mohana

Mohana Ravindranath is a Bay Area correspondent covering health tech at STAT and has made it her mission to separate out hype from reality in health care.