Welcome to This, er, Summer in Digital Health. We’re going to pretend I took the summer off from highlighting key headlines in healthcare technology and why they matter because I was busy peering off into the distance contemplating the right format to better engage with my readers and not because the least stressful way for the Beastwoods to navigate the transition from preschool to summer camp to summer vacation to kindergarten was for me to work quite a bit less.

Anyway … welcome to my new newsletter format! There’s still a handful of key themes, though there will probably be fewer than I’ve featured in the past. What’s new? More insights straight from my brain! When lots of headlines drop on the same topic, it’s easy to miss the forest for the trees. Here’s to hoping my perspective / free time to actually sit with these stories helps you make better sense of it all.

Does healthcare fear AI is too risky for clinical care?

If the first few months of 2026 were about the promise of AI for clinical care, as seemingly every tech company released an AI agent to automate this and streamline that, the headlines this summer seemed to suggest the honeymoon period is over.

• Patients in rural areas are wary of AI, which many states are pushing as part of their Rural Health Transformation efforts.
• More than 70% of Americans want to know when providers use AI as part of care delivery. That’s important because…
• …a similar percentage of provider organizations use AI that hasn’t gone through a formal review process, meaning that IT teams haven’t technically approved it.
• Organizations have also underestimated the malpractice risk of AI scribes that could introduce errors in records or simply produce poor-quality notes.
• There’s also evidence to suggest AI will increase healthcare costs in the short term, even as it increases access to care, because technology implementation and workflow integration aren’t exactly easy.
• But hey, a bunch of guys with stakes in AI companies think autonomous AI will replace physicians in a few years.

My take: AI for clinical care is starting to seem a lot like coffee. One day, it’s bad for you. The next day, it’s good. The next day, it’s still good, but only in excessive quantities that make you jittery, paranoid, and irregular. Organizations clearly know they need to erect guardrails and draft governance policies, but FOMO seems to get in the way of taking these pragmatic and logical but also time-consuming steps. Unfortunately, that inaction is going to bite them back. I wouldn’t be surprised if the majority of pilots die on the vines in the next several months, along with the companies that created them.

With wearable data, everything old is new again

This summer, both Apple and Google announced updates to the health, fitness, and safety features linked to their watches and wearables. Meanwhile, a recent AMA survey found that most physicians do in fact review wearable data, 77% find it valuable for clinical care, and close to 30% have acted on the data they saw – findings that even surprised the association’s CEO.

However, only 6% have been able to integrate wearable data with clinical workflows. On top of that, many physicians don’t really trust wearable data, especially if it comes from FDA-cleared and not FDA-approved devices. Plus, few bother to use diagnostic and billing codes for remote monitoring, which would be the main motivation to look at and act upon wearable (or sensor) data in the first place.

My take: The challenge of figuring out what clinical teams are supposed to do with heaps of consumer wearable data is a tale as old as time. The evolution of technology over the last couple decades – from spreadsheets to databases to dashboards to generative AI summaries – clearly hasn’t made things any easier. As with a lot of healthcare’s problems, this one would have been solved by now if stakeholders wanted it to be. Since there’s no incentive for providers to even look at data, let alone interpret it and incorporate it into care plans, it’s just going to sit there, largely ignored, as it always has been.

Digital health apps benefit some conditions more than others

A literature review in The Lancet found that digital health apps work well for managing many conditions and symptoms – namely blood pressure, blood glucose, mental health, body weight, physical activity, and medication adherence – but not everything. Managing cholesterol, body fat, heart disease, and healthy diets proved to be difficult with apps alone, as these areas often require clinical inventions in addition to the educational content, feedback, and support loops that apps tend to specialize in.

My take: A lot of healthcare research confirms things we already knew, or at least suspected to be true. While this paper fits that description, that’s not necessarily a bad thing. Given the industry’s incessant push to do more with less, particularly when it comes to engaging with patients using technology that never quite lives up to its ROI potential, it’s helpful to have clear evidence of which apps can stand alone and which apps would need an all-important human touch and, depending on who, you ask, may not provide enough juice to be worth the squeeze.

The Epic effect remains strong but shows some cracks

Research firm Redesign Health published a report on the Epic Effect, or the strong gravitational pull of the industry’s leading technology vendor. More than 70% of health systems surveyed described themselves as Epic-first, and more than 90% are confident the vendor can execute on its AI roadmap. That said, nearly two-thirds would buy from a startup if it did something Epic doesn’t do as well, such as imaging, quality reporting, and scheduling and discharge. Also, 48% of health system admit they select Epic because it’s “good enough,” which the consultancy said suggests that “Epic gets to clear a lower bar.”

My take: Winston Churchill is credited with saying that democracy is the worst form of government except for all the other ones. Healthcare seems to feel the same way about Epic. Keynote speakers and other thought leaders are quick to offer criticism – some of it warranted, given its history of integration frustration and recent antitrust allegations – yet the company’s market share is only going up. Yes, there are sunk costs associated with keeping your current EHR, and it sure helps when the same vendor offers integrated features that meet needs you didn’t realize you had. At the same time, let’s remember that Judy founded Epic in 1979 (the year before I was born, mind you). That means the industry has had almost five decades to build a better alternative.

Bulking up build efforts in the build vs. buy debate

A couple publications offered new takes on the omnipresent build vs. buy question for healthcare’s technology teams. Healthcare Dive unpacked the effort to build oncology patient management tools at NYU Langone Health and Dana-Farber Cancer Institute, and Digital Health Insights explained the Cedars-Sinai approach to engage stakeholders from the beginning to ensure efforts succeed. That engagement helps avoid the pilot purgatory problem that plagues innovation efforts viewed less as tried-and-true technology implementations and more as experiments; it also helps demonstrate to decision-makers that teams know what they’re doing and deserve some leeway to tinker if they have a good idea.

My take: It’s interesting to see this topic continually resurface, as it’s a tale almost as old as time. For general-purpose technology, or anything that needs to be done at scale, Buy is a pretty clear answer – or Partner, if a health system is large enough to meaningfully influence what a technology vendor will do. Build makes more sense for anything highly specialized and/or personalized, either because the patient (sub)population is unique or the process is nuanced (a polite way to say complicated). Yes, AI is moving the needle on scaling specialized, personalized, and nuanced workflows enough to shift some technology from Build into Partner or Buy, but systems still need to weigh how much control over the process they’re willing to give up for what very well may be the temporary convenience of not doing it themselves.

Digital health funding trends don’t shock or surprise

The latest analysis of digital health funding from Rock Health found that deals are getting a little bit bigger, while deals more than $100 million make up an increasing volume of all capital invested. The firm found that investors prioritize domain expertise, scalability, post-implementation support (especially when it comes to AI), and broad partner ecosystems. Conversely, Modern Healthcare reported that investors shy away from point solutions, products that don’t scale, and business models solely reliant on reimbursement models subject to the whims of the Trump administration (my words).

My take: The points about which product categories aren’t seeing much investment make perfect sense and suggest the digital health landscape is maturing quite a bit. This also ties back to a couple of the other themes. If Epic is going all-in on AI, it stands to reason that AI startups will soon face a steep uphill battle and cannot rely on their products alone. If more health systems are comfortable rolling their own software, there’s less room for unfamiliar partners. Plus, just about everyone’s operating on razor-thin margins, and ROI is one of the first terms to come up in any conversation about technology investment. I get the feeling a slowdown is coming.