This Month (or So) in Digital Health: AI’s Growing Pains, Federal Health’s Shrinking Pains

Welcome to This Month in Digital Health, where I summarize news articles that recently caught my attention and attempt to try to put the trends in broader context. Stop me if you’ve heard this before, but one of the main themes this month is AI, which healthcare finally seems to be putting under the microscope. Also of note is how the Iran war will impact health systems and how federal health may be gutted yet again.

The honeymoon’s over for clinical AI. One study found clinicians save less than 30 minutes a day using AI scribes and don’t decrease their after-hours EHR use. Another one determined Epic’s algorithms don’t perform well in the real world, which matters because AI vendors struggle to compete with EHR incumbents even if their products work better. So, it comes as no surprise that healthcare’s struggling to calculate AI’s ROI and getting impatient when ROI is slow to arrive. On top of everything else, experts say AI oversight and governance need to catch up. Good times.

Consumers understand AI’s limitations. About one-third of American adults turn to AI with health-related questions. Don’t run around screaming that the robot doctors are taking over, though. Education and research make up more than half of healthcare AI queries, compared to 8% for care questions, and people readily admit they use AI because it’s convenient, not because it’s accurate. Seems like there’s most definitely some log-hanging fruit for all those consumer AI health tools (nearly diagramed by Rock Health).

Federal health funding? What federal health funding? The White House is proposing a 12% cut to HHS in 2027, though it’s worth noting Congress ignored a similar request last year. Even with funding restored, HHS is in disarray after massive job cuts (as are other federal agencies), as detailed in excellent long reads from Healthcare Dive. Also, 400-plus hospitals are at risk of closing due to Medicaid cuts that are projected to reach $1 trillion over the next decade.

Healthcare may feel the pinch of the Iran war. Disrupted shipping in the Strait of Hormuz may not impact medical supply chains immediately, and the uncertainty of what may come means stocking up is a bad idea. Cybersecurity is a more immediate threat, as “cyber conflict increasingly mirrors geopolitical tensions,” one expert told Healthcare Brew.

Prior authorization: Now with slightly less time wasted! The insurance industry patted itself on the back by eliminating 11% of prior authorizations, according to an AHIP statement. Meanwhile, Health Affairs found 90% of prior authorizations are approved the first time, which essentially means the $30 cost per transaction is paying for a rubber stamp.

In other relatively interesting news:

Thanks for playing along. See you next time.

This Month in Digital Health: Beyond the Conversations With Well-Dressed People in Cushy Chairs

Welcome to This Month in Digital Health, where I summarize articles that recently caught my eye to explain why I think they’re important. With ViVE and HIMSS behind us, I figured I’d shy away from all the product and partnership announcements that made headlines, which explains why this is a bit shorter than usual. There were still some compelling stories out there, from the usual AI and rural health challenges to ruminations about how seemingly competing health data exchange efforts will in fact complement each other.

AI: Everything everywhere all at once at the same time. A Health Affairs paper argued healthcare needs its AI bubble to burst, as organizations are chasing “innovation without substance.” It doesn’t help that large language models are susceptible to misinformation (just like people), health systems are still wrestling with human-in-the-loop approaches (focusing too much on individual outputs and not enough on scalable frameworks), and LLMs are impacting how people interact with search results (fewer clicks on reputable hospital websites, which of course isn’t a good thing). If nothing else, I suppose, robots can run the hospital cafeteria.

Rural health: Because healthcare loooooves to rush into things. KFF Health News is all over rural health transformation, where proposals range from investing most of the money (Wyoming) to using robots in maternity care (Alabama). Plus, apparently there’s some tension about plans that were drafted and approved, in no small part because states didn’t really have a lot of time to come up plans. That very well may have been the point – I’ve heard folks argue that CMS was looking to fund projects that were well past the planning stage – but the whole thing’s starting to look rather messy, if not ill-conceived.

CMS: TEFCA and the Health Tech Ecosystem are totally besties. It’s not the best look when you launch a data-sharing initiative and then have to come out and say it’s not competing with an existing data-sharing initiative with similar characteristics and motivations, as CMS and ASTP had to do regarding TEFCA and the Health Tech Ecosystem. Apparently, the ecosystem is an “accelerator,” and its work may (or may not) be folded into TEFCA at some point. Glad we cleared that all up.

ACCESS: With rates this low, who needs a doctor? The proposed annual rates for the ACCESS model (Advancing Chronic Care with Effective, Scalable Solutions) are, um, well, they’re not very high at all. A lot of folks are arguing that’s exactly the point, as the model’s meant to attract 1) entities that take a digital-first approach to chronic care and 2) patients that don’t necessarily need a lot of in-person care. I get it, but I’m not sure going out of your way to alienate physicians is really the best approach.

In other news:

See you next month, when there will be 95% fewer healthcare IT articles featuring photos of well-dressed people sitting in comfy chairs and holding microphones.

This Month in Digital Health: Some things are actually looking up!

Welcome to This Month in Digital Health, where I highlight articles and trends that recently caught my eye and attempt to explain why I think they matter. This time around, there’s actually maybe possibly some decent news about chronic care management and care at home, though rural health, AI, and support for the nursing workforce are all a bit messy.

AI is still complicated. HHS is setting a course to broaden AI adoption internally, with the hope of setting an example for the private sector to follow. It seems that healthcare organizations may benefit from any guidance they can get: They’re struggling with AI governance, feel underprepared for AI deployment, aren’t getting AI ROI fast enough, and aren’t yet seeing major productivity gains from AI.

Chronic care is getting good attention. CMS announced the ACCESS model for pay providers for using tech to support chronic disease management. Many see it as a “bold new model” for tech-enabled care, and vendors see it as an opportunity to prove their worth to providers who are typically skeptical of such things. That said, providers have a lot to consider when it comes to who to enroll, what tech to use, and how program structure ay change.

Finding rural health details the hard way. CMS isn’t talking about what states plan to do with Rural Health Transformation money, so KFF Health News filed public records requests and reported on the details. There’s a lot about improving access to food, medication, and lifestyle improvements – just not specifically for Native American tribes, who weren’t eligible to apply and have to rely on their state governments.

More care at home momentum? Former CDC director Susan Monarez penned a piece for Chief Healthcare Executive suggesting up to 90% of care can take place in the home. A JAMA Network Open study found care at home is especially effective in rural areas, which bodes well for programs looking to demonstrate value. Speaking of which, vendors are pushing for value-based contracts from Medicare, which would help with their own long-term stability.

The balancing act for supporting nursing teams. An interview with the Mayo Clinic in HealthTech Magazine (a client of mine, for the record) highlighted the benefits of AI to support nursing workflows – provided that nursing teams get a say from the get-go in what tools will do and how they’ll be implemented. Saving time should be a point of emphasis, as nurses who skip breaks or stay late are probably burning the candle at both ends.

Meanwhile, I found a bunch of odds and ends this month.

Happy Holidays. If you have children home from school for the next couple weeks, remember that no one’s going to judge you if they wear pajamas all day, eat candy for breakfast, and use every pillow and couch cushion in the house to make an obstacle course or fort – especially if you don’t tell anyone about it…