Welcome to the eighth edition of This Month in Digital Health, where you get to read my very own thoughts on the news and trends that caught my attention over the last few weeks. We’re a bit tardy this month – Sandwich Generation Caregiving for the win, am I right? – which means there’s been extra time to gather hot takes on AI, data, healthcare costs, and GLP-1s. Have at it, folks.
AI likes it easy. Healthcare isn’t. Therein lies the rub. Yet another paper has shown that LLMs can outperform physicians when making rote decisions (even general-purpose LLMs) that don’t have to account for the stuff that happens in real life – whether it’s delays in prior authorization or claims denials powered by AI or, like, patients’ unaddressed social needs or the stress of the hospital. Nor does it help that providers fear overreliance on AI will diminish clinicians’ critical thinking capabilities.
Who can access data? What is data? What is access? Rock Health found that 57% of Americans own at least one wearable, and many are tracking health metrics. That’s good! The bad news: Wellness apps interpreting device data aren’t really regulated, and physicians have no idea how they conclude a patient is at risk of, say, depression. Plugging data into an LLM isn’t much better, as general-purpose LLMs aren’t HIPAA covered entities (though the healthcare-specific ones are).
GLP-1s are taking over healthcare, and not in the best ways. Speaking of wearables, Mass General and Samsung Health are partnering to see whether a smartwatch can track muscle loss in adults on a GLP-1. That’s good, because it’s a common concern for patients who lose weight. Other concerns are emerging, too, including GLP-1 prescriptions from telehealth providers who don’t talk about the risks or the importance of healthy habits, and the efficacy of GLP-1 pills, which has self-employed insurers thinking twice about insurance coverage.
Great googily moogily, healthcare is expensive. On top of everything else, GLP-1s are making healthcare even more expensive – to the tune of 9% increases in commercial costs in 2027, per PwC. Only 7 nonprofit Blues plans had positive operating margins in 2025, and many provider-sponsored health plans are giving up amid rising costs and general industry instability. Speaking of which, more than 75% of executives think healthcare’s financial instability will reach an existential tipping point before the end of the decade. Good times.
Meanwhile, in other stories of interest:
- Given the odd juxtaposition of primary care (with low costs and generally stable utilization) and health insurance (which is meant to cover expensive, unpredictable events), a JAMA paper proposed the notion of primary care as a public utility through which states create a common fund that pays practices directly. Insurers could still compete on other lines of business. I’ve heard worse.
- The ACCESS Model for Medicare goes into effect on July 5. With 150+ participants already lined up, Second Opinion posed a reasonable question: What about ACCESS for Medicaid? Makes sense to me, especially since states have embraced managed care.
- When it comes to rural health transformation, the numbers don’t add up. As $50 billion in federal aid doesn’t equal nearly $1 trillion in losses, states are focusing on cost-cutting strategies, not innovation. Good job, everyone.
Happy summer, Northern Hemisphere. If you know me, you know this is my least favorite time of the year. (And if you don’t, well, now you do.) If you need me, I’ll be under an oscillating fan, within a foot of an AC, and waiting until the humidity breaks to do anything that will cause me to begin to sweat.