 | ● |  |
|
|
|
| | | News and views you ought to know about:- Folks in at least 1 of the 6 states in which Medicare will pilot AI for prior authorization don’t know why their corner of the U.S. was chosen. CMS announced the program in June, but many Washington Staters are only now finding out they’ll be affected when the automation program goes live Jan. 1. “This just is unfortunate and surprising news,” Matt Hollon, MD, MPH, remarks in the Washington State Standard Sept. 25. “Patients expect their care to be guided by doctors, not insurance companies or automated systems.”
- Hollon is a primary care physician and university professor who serves as a regent with the American College of Physicians and a VP with the Washington State Medical Association. If he’s surprised, imagine what less-connected colleagues must be thinking. And that’s to say nothing of the Evergreen State’s cohort of older patients. Washington has about 1.6 million Medicare beneficiaries, about half of whom are enrolled in traditional Medicare and the other half in Medicare Advantage, the Standard notes.
- CMS tells the outlet the state was selected as a co-pilot locale to help the agency test “across diverse practice environments to ensure a reliable and valid model test.”
- Recall that the test model, known as Wasteful and Inappropriate Services Reduction, or WISeR, will run through 2031. The other five states are Arizona, Ohio, Oklahoma, New Jersey and Texas. The program will contract with third-party companies to deploy the prior-auth AI algorithms. Wisely, WISeR calls for a human in the loop to review every claim for which the AI recommends denial.
- Last month a group of Democrats in the U.S. House quizzed CMS Administrator Mehmet Oz, MD, on 13 aspects of the WISeR pilot. At the top of their list: What criteria were used to select the six states for this model? As of Sept. 22, they hadn’t heard back from Dr. Oz. “This should be a bipartisan issue,” U.S. Congresswoman Suzan DelBene, D-WA, remarked to the Standard. “The states are across the country. There should be concern about making sure that people have access to care in all these states and understand exactly how the models work, why it’s set up this way, what their goals are.”
- The outlet further reports that some WISeR critics “see the initiative as a sign of the Trump administration creeping toward Medicare privatization” and are “unsettled by the use of AI.” Read the whole thing.
- Industry healthtech romances can turn nasty, and fast, when one of the parties starts to outgrow the relationship. Two major companies may be about to learn this the hard way. If they do, both will endure the indignity of having a messy breakup play out in public. The parties are the needs-no-introduction Epic and the healthcare AI startup Abridge, which is hot and getting hotter. In the business sense, of course. “Epic’s push into AI highlights a larger tension that’s playing out across tech, mirrored in relationships from Microsoft-OpenAI to Google-Anthropic,” notes Business Insider in coverage of the suddenly rocky relationship. “In the fast-moving AI race, what happens when dominant incumbents morph from partners and shareholders into competitors?”
- It’s hard to know whether the once-cozy partners are fighting like cat and dog or just entering a tentative let’s-stay-together-for-the-kids stage, suggests an Abridge investor who spoke on condition of anonymity. “They’re still talking to each other,” the source tells BI, “but there are a lot of things that make you feel like, wait, what? What’s going on here? It’s a complicated picture.”
- The reportage may sound gossipy, but it’s more instructive than voyeuristic. Up to this juncture, anyway. Get the rest of the story.
- Modern careers in digital healthcare technology are not only mentally stimulating and morally rewarding. They pay quite well too. That’s one takeaway to be gleaned from an information item posted by the University of Central Florida. And just because the school uses the page to recruit new students doesn’t mean it’s not of interest to a wider audience. As UCF has it, healthcare IT project managers make between $90K and $130K a year while healthcare cybersecurity specialists earn as much as $140K. That’s not exactly competitive with C-suite compensation, but it’s nothing to yawn off, either. And if you’re talented, dedicated and business-oriented, you can always work your way up the org chart. “Since each career requires a unique set of skills and educational backgrounds,” UCF notes, “students interested in pursuing a HealthTech career should consider obtaining a degree in health information management, biomedical engineering or computer science.”
- Radiologists have been using AI to outline areas of clinical interest in imaging scans—a process called segmentation—for some time. Researchers at MIT are taking the AI piece further. Their innovation, which they’re calling MultiverSeg, predicts segmentation based on user interactions such as scribbles made by very human hands. And it does this while contextualizing each segmented image such that the algorithm can refer to the area of interest later, as appropriate. “When the user uploads a new image and marks areas of interest,” MIT’s news operation reports, “the model draws on the examples in its context set to make a more accurate prediction”—and with less user input, at that. What’s more, the user continuously refines the AI’s performance just by interacting with it, explains an MIT computer-science grad who lead-authored a paper on the tool. This automated finetuning, she explains, “dramatically accelerates the process because it is usually faster to correct something that exists than to start from scratch.” Learn more.
- Now cometh a robot surgeon that—or who—can operate on patients solo. As in with no human in the room. No one is suggesting the droid will do so anytime soon. But it’s bracing enough just to know that, according to its developers, it could. Especially since they’re with one of the most prestigious academic medical institutions in the world. Friends, we’re talking Johns Hopkins. “This work represents a major leap from prior efforts because it tackles some of the fundamental barriers to deploying autonomous surgical robots in the real world,” says the lead author of a scientific paper describing the work, Ji Woong “Brian” Kim (who has now moved on to Stanford). “Our work shows that AI models can be made reliable enough for surgical autonomy—something that once felt far-off but is now demonstrably viable.” The Johns Hopkins news team has coverage.
- Also worthwhile:
- From AIin.Healthcare’s sibling news outlets:
|
| | | |
| |
| | | Is Your AI Notetaker Putting You at Risk? AI meeting tools can quietly introduce security and compliance risks when they record or store conversations without the right safeguards. In healthcare, that means potential exposure of PHI and costly governance gaps. This free Security Checklist from Fellow gives IT, Ops, and compliance leaders seven simple checks to evaluate any AI notetaker against standards like HIPAA, SOC 2, and ISO 27001. Use it to spot red flags early and ensure your organization stays protected. Before Shadow AI spreads inside your org, download the checklist to reduce your risk here.
Join Us at the Global Summit Driving Systemic Solutions to Clinician Burnout Nabla is partnering with the Ending Clinician Burnout Global Summit 2025 (Nov 6–7, virtual), a global community working together to reimagine healthcare and build lasting solutions to the clinician burnout crisis. This two-day event, co-hosted by Ending Clinician Burnout Community and Sharp Index, will bring together leading institutions like Mayo Clinic, Duke Health, Yale Medicine, Johns Hopkins, and Stanford, alongside frontline clinicians, to design systemic changes that protect clinicians and strengthen patient care. 👉 Learn more and register here: endingclinicianburnout.com
Major breakthrough in AI medical transcription
Speechmatics’ new medical model hits 93% accuracy and cuts overall word errors by 17% vs. the next best, halving keyword errors on medical terms. Which means first-pass notes that hold under noise, accents, and overlap, with 96% keyword recall and a 4% keyword error rate. Run it live or in batch. Learn more.
|
| | | |
|
| |
|
|
