Five AI Signals from HIMSS26 That Will Shape the Next Era of Healthcare
This year at HIMSS26, the conversation around AI felt meaningfully different.
In past years, much of the focus was on what AI might do for healthcare in terms of how it could reduce administrative burden, improve diagnostics, or unlock new insights from clinical data. Walking the floor and speaking with providers, technology leaders, and policymakers this year, I heard a different set of questions.
The conversation has shifted to practicality and whether AI can deliver real value in the environments where care actually happens.
A few clear points stood out to me.
1 - AI is moving into a phase of production and deployment
One of the most noticeable changes at HIMSS this year was the shift from pilot programs to production systems.
Over the past few years, many health systems have been experimenting with AI. But what I heard repeatedly in conversations last month is that experimentation is no longer enough, and health systems are asking how something will actually work in practice.
They want to know whether AI can reduce clinician workload, improve decision-making, and produce measurable outcomes. And if it can’t demonstrate clear ROI, it’s not going to move forward.
To me, this signals a turning point where AI is quickly becoming part of the infrastructure (and being evaluated accordingly).
2 - Interoperability alone isn’t enough
Interoperability has been a major focus in healthcare for years, and there’s been real progress. But one thing that became very clear at HIMSS26 is that moving data is not the same as making it usable.
We now have more pathways to exchange data than ever before, but much of that data is still fragmented, inconsistent, and difficult to interpret in a clinical context.
In my view, the industry is starting to recognize that interoperability is only the first step, and the real challenge is turning that data into insights that reflect the patient’s full story in a structured, reliable way.
Without that layer of transformation, data exchange alone doesn’t unlock the value that AI promises.
3 - AI needs the whole patient record
Another theme I heard repeatedly is that AI in healthcare cannot rely only on structured data.
Some of the most important clinical information still lives in unstructured formats, such as physician notes, pathology reports, radiology summaries, and historical records that were never designed for machine use. That’s where much of the nuance is.
From my perspective, this is one of the biggest gaps in current AI implementations. If you’re only looking at structured fields, you’re often missing critical context that influences care decisions.
This becomes especially important in complex areas like oncology, where patient histories are long and highly individualized. To make AI truly useful in these settings, you have to be able to work with the full patient record.
4 - Trust, governance, and traceability matter more than ever
As AI moves closer to clinical decision-making, trust becomes non-negotiable.
One of the more consistent themes at HIMSS26 was the growing emphasis on governance and oversight. Health systems are becoming much more rigorous in how they evaluate AI tools, especially in terms of transparency.
What I’m seeing is a shift toward questioning whether clinicians can trace and verify the sources of key insights, which is critical given the real consequences of healthcare decisions. If AI is going to play a role in those decisions, clinicians need to be able to trust AI systems and ground that trust in traceability and data integrity.
I sense this is where a lot of AI efforts will either succeed or fall short.
5 - Implementing AI is a significant challenge
If there’s one overarching takeaway I had from HIMSS26, it’s that the biggest challenge is implementing AI systems in clinical settings.
We already have powerful AI capabilities, so the harder problem is integrating those capabilities into clinical workflows in a way that helps providers.
What I heard consistently is that AI needs to meet clinicians where they are. It has to fit into existing systems, surface insights at the right time, and truly reduce friction.
That’s a much more complex problem than developing a model in isolation. It requires a deep understanding of how care is delivered and how clinicians interact with information in real time.
My big takeaway: The next era of healthcare AI will be Built on data infrastructure
Stepping back, the signals from HIMSS26 point to a broader shift in how we think about AI in healthcare.
We’ve moved on from obsessing over the sophistication of frontier models and are focused on whether the underlying data that informs AI platforms is complete, reliable, and usable in a clinical context.
It’s likely that the next era of healthcare AI will be defined by the ability to turn fragmented patient data into trusted, decision-ready intelligence.
That’s what will determine whether AI can move from promise to practice, and ultimately, that’s what will determine whether it can make a meaningful difference in patient care.