Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Physician in the Loop

Physician in the Loop artwork

Daily Briefing for November 1, 2026

Nov 1, 20266 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Affordable Care Act premiums rise another 13% as sign-ups open The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source. Read the post and its sources Get the free email briefing Average benchmark premiums for Affordable Care Act marketplace plans are rising 13% for 2027, KFF reported, as open enrollment began Sunday after a 26% increase for 2026. The two increases add up to about 42% since 2025, according to KFF's analysis, which attributes the rise to higher underlying care costs and changes in the risk pool after enhanced premium tax credits expired. Enrollment on HealthCare.gov runs through Jan. 15, 2027, the Centers for Medicare & Medicaid Services said, and the Congressional Budget Office projects marketplace enrollment will fall to 13.1 million in 2027 from a 2025 peak of 22.4 million. Separately, a Doximity analysis of more than 8,500 dosing questions found general-purpose AI models gave seriously wrong or potentially lethal doses in about 2% of answers, and fewer employers are paying for GLP-1 weight-loss drugs, a WTW survey found. In this episode Affordable Care Act benchmark premiums rise 13% for 2027 after a 26% jump this year, as enrollment opens Nov. 1 General-purpose AI models gave seriously wrong drug doses in about 2 of every 100 answers, Doximity analysis reports AI reading 25,000 primary care notes a week found about 7 times as many missed specialist referrals at Singapore clinics Employer coverage of GLP-1 weight-loss drugs falls to 53%, and 30% of employers still paying are weighing dropping it Penn Medicine AI trained on 400,000 CT images screens for hundreds of conditions a scan was not ordered to find

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show.

Voice 1: This is Physician in the Loop, the daily briefing on AI in medicine. It's Sunday, November 1st, 2026.

Voice 1: Premiums on the Affordable Care Act marketplaces are rising 13% for 2027. That is on top of a 26% jump this year. KFF reported the finalized rates on October 30th.

Voice 2: 13% on top of 26%. Does that land the same everywhere?

Voice 1: No. It is 11% in the states that use HealthCare.gov and 16% in the state-run marketplaces. KFF says the driver is higher underlying care costs plus a changed risk pool, now that the enhanced premium tax credits have expired. Anyone above 400% of the federal poverty level gets no credit at all anymore.

Voice 2: Is enrollment actually holding up under that?

Voice 1: It is dropping. Paid enrollment in February was down 12% from a year earlier. And insurers are leaving. At least 11 had announced exits from a state as of October 29th, including Cigna, which is pulling out of every one of the 11 states where it sells individual plans.

Voice 2: So people are shopping a smaller market with fewer insurers and a much higher price tag.

Voice 1: Open enrollment on HealthCare.gov runs now through January 15th, CMS says. That is the window physicians' uninsured patients are working with this year.

Voice 2: A very different kind of safety number came out of Doximity this week.

Voice 1: General-purpose AI models gave a seriously wrong or potentially lethal drug dose in about 2 of every 100 answers. That is from a Doximity analysis of more than 8,500 clinical dosing questions covering more than 2,200 FDA-approved drugs.

Voice 2: What kind of mistake was most common?

Voice 1: Failing to adjust the dose for age, kidney function or drug interactions, even when the question handed the model that information directly. And newly approved drugs had five times the odds of an error.

Voice 2: Doximity has a financial stake in this story though, right?

Voice 1: It does. They released this alongside their own dose-checking tool, which they say is correct nearly all the time. These are the company's own figures, the report did not name which models were tested, and none of this is peer reviewed.

Voice 2: Worth hearing, with that asterisk attached.

Voice 1: A quieter AI story out of Singapore is about catching things that get missed, not errors that get made.

Voice 2: An AI system scanning consultation notes at seven National University Polyclinics sites found missed specialist referrals about seven times more often than before. That is in a study published in npj Digital Medicine.

Voice 1: Seven times as many catches. Did clinicians agree with what it flagged?

Voice 2: The post doesn't repeat the full confirmation numbers, but in a blinded one-week comparison of 83 flagged cases, the system's positive predictive value beat the confirmation rate from seven physician reviewers, 62% against 52%. It also ran with very high specificity, though sensitivity was under half.

Voice 1: So it is better at not crying wolf than it is at catching every missed referral.

Voice 2: Right. The study also projects about 41 fewer staff-hours, but reports no actual patient outcomes yet.

Voice 1: A money story on the other end of the spectrum: employers are backing off GLP-1 coverage.

Voice 2: Employer coverage of GLP-1s for weight loss fell to 53%, down from 57% a year ago. WTW surveyed 471 employers with at least 100 workers for its 2026 Best Practices report.

Voice 1: First decline in three years. What's happening among the employers who still cover it?

Voice 2: 30% of them are weighing dropping it. And among employers that already dropped it or are considering dropping it, 91% blamed cost.

Voice 1: That's a lot of agreement on the reason. What's the overall cost picture look like for next year?

Voice 2: Employers project median health costs will rise nearly 10% in 2027 even after they make changes to their plans. That would be the highest rate since the early 2000s. WTW's Jeff Levin-Scherz put it this way: pharmacy has moved from a line item to the center of the healthcare cost conversation.

Voice 1: A patient losing coverage mid-prescription is a very different conversation than a line item.

Voice 2: A last story, and it's a scan looking for things nobody asked it to look for.

Voice 1: Penn Medicine researchers built an AI model called Percival that runs quietly on chest and abdominal CT scans, hunting for disease beyond whatever the scan was actually ordered to check. It is described in a study in npj Digital Medicine.

Voice 2: Trained on what, exactly?

Voice 1: More than 400,000 CT images from nearly 52,000 patients in the Penn Medicine BioBank. It beat two other AI imaging models across 279 chest conditions and 230 abdomen and pelvis conditions, with its biggest edge on lung cancer and acute lymphoid leukemia.

Voice 2: No accuracy numbers attached to any of that?

Voice 1: None given in the report. It is still in preclinical testing, hasn't been validated outside Penn Medicine, and developers hope to start testing in early 2027.

Voice 2: So a scan ordered for one thing could someday flag something else entirely, assuming it holds up outside one health system.

Voice 1: Before we go, something to watch. Senator Bill Cassidy released a discussion draft of the PATIENT AI Act on October 29th.

Voice 2: What would it actually do?

Voice 1: It would direct HHS to set a coverage framework and have the CMS Innovation Center test a payment model for autonomous AI. Becker's reported comments are due December 1st.

Voice 2: A real payment pathway for AI, not just a pilot buried somewhere.

Voice 1: That's the idea, though it's only a draft so far. Separately, Palmetto GBA and Noridian have proposed dropping coverage for AI brain MRI quantification, with comments open through November 14th.

Voice 2: One agency pathway opening for AI payment while two contractors are moving to close one down. That's the kind of split a hospital's billing office is going to be untangling for a while.

Voice 1: Sources and corrections are at physicianintheloop.org, where you can sign up for the free email briefing. This briefing was read by AI voices.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Special Topic: What changes on Jan. 1

    Nov 20, 20267 minAudio

    Laws in at least eight states that touch AI in health care take effect on Jan. 1. Most bind insurers and chatbot companies. A few, in California and Colorado, reach the practice itself. Read the special topic and its...

    Credit the people on this episode

  2. Friday Letter: On a short leash

    Nov 20, 20267 minAudio

    Software has begun to act for doctors, not just advise them. It should be kept on a short leash, because the doctor watching it rarely notices when it slips. Read the letter and its sources:...

    Credit the people on this episode

  3. Daily Briefing for November 20, 2026

    Nov 20, 20266 minAudio

    Hospitals race to patch a Java flaw found with AI in an hour The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source. Read...

    Credit the people on this episode

  4. Special Topic: Not the doctor's to sign

    Nov 19, 20267 minAudio

    Free AI tools for doctors come with a HIPAA contract one doctor can accept alone. A hospital's records need the hospital's signature, and the companies should make the safe path the default. Read the special topic and...

    Credit the people on this episode

  5. Daily Briefing for November 19, 2026

    Nov 19, 20267 minAudio

    AI hacking agents seize some networks in under 3 hours, FBI warns The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source....

    Credit the people on this episode