Voices of U of U Health
Becoming an AI-Forward Health System
By Ken Kawamoto, MD, PhD, MHS, FACMI, FAMIA
I came across a statistic years ago that really stuck with me. People in the United States only receive about half the care they should. The clinical guidance and data are available. We have care teams ready to help. But what’s usually missing are the tools to optimally support that care.
This challenge is what pulled me into the field. It’s one many of us have struggled for years to address at scale. Now, I think we’re finally at a point where we can address such gaps in care in a systematic way.
I Didn’t Set Out to Work In AI
My early training pointed me toward basic science. However, that changed when I started caring for patients. What really stood out wasn’t just the complexity of medicine. It was how little support clinicians had when they were overwhelmed with information, short on time, and making high-stakes decisions. That problem stuck with me and pulled me into informatics, where it’s shaped my work ever since.
For years, I’ve asked, “How do we give clinicians and care teams better support when they need it?” A lot of that work has centered on better integration, processes, and data use through efforts like ReImagine EHR. Our goal has always been to use technology not for its own sake but to make care safer, more effective, and more compassionate.
Why This Moment Feels Different
I was an AI skeptic for a long time. While tools like early versions of ChatGPT seemed really promising, they often fell short when we rigorously tested them.
What has changed is that the technology has improved so much, so quickly. Tasks that once felt out of reach are now possible in practical ways. AI can now sum up a patient’s story when their chart is overwhelming. It can help write up notes, so clinicians spend less time writing and more time thinking. It can help identify important details sooner. Now, these tools are quite reliable for clinical care when configured and deployed appropriately.
From Vision to Practice
With the framework of Vision 2030 in mind, Bob Carter, MD, PhD, CEO of University of Utah Health, has challenged us to think about what it means to become an AI-forward organization.
To me, that doesn’t mean chasing every new product or trend. It means being thoughtful, strategic, and patient-focused about where AI can really make a difference. It means using these tools to help improve care for patients while making work more manageable for our care teams.
It’s the reason for the Innovation Lab at University of Utah Health, which is one of my key areas of focus as our health system’s inaugural Chief Health AI Transformation Officer. The lab started in January 2026 as a fast-paced center for new ideas focused on our clinical, research, and education missions. It was built to quickly turn promising ideas into working models, then see what works and expand it safely.
In other words, the lab isn’t just a place to explore what AI might do someday in the future. It’s a place to test what a health system that embraces AI can do right now.
What Does This Look Like in Everyday Use?
Early projects already show how AI can make day-to-day care better. For example:
- Identifying Heal at Home candidates. Some patients can safely continue care at home rather than staying in a hospital bed. Working with the leaders of this initiative, we now have tools that monitor patients in real-time and help identify who may be a good fit for Heal at Home.
- Clinical trial matching. AI can help screen patient populations against trial criteria and bring possible matches into view faster, which may help more patients get connected to treatment options. We are currently piloting this approach with Huntsman Cancer Institute leaders.
- AI Workbench. One of the ideas we’re most excited about is establishing a safe, self-service environment where frontline users can build and share their own AI tools for improving patient care inside our Epic electronic health record system. The people closest to the work usually know where the friction is. They should be empowered to leverage the latest AI technologies to help solve it. We are currently in pilot use of this tool, and it’s looking really promising.
That, to me, is what “AI-forward” should mean. Each project starts with a real problem in health care and asks whether AI can make things easier in a way that’s helpful and safe.
Safe By Design
As encouraged as I am by what we’re building, safety has to stay central. That’s why, for example, we’re particularly cautious about tools designed for use directly by patients, and we’re building in processes and technologies to help ensure safety. If something makes work faster but less safe, that’s not a win.
Also, the goal isn’t to replace our physicians, nurses, or clinical staff. The point is to support them better. Ultimately, a health system that is serious about AI isn’t just adopting the newest or fastest tools. It’s one that’s earning trust by using them safely and responsibly so there’s more time to spend with patients.
What Comes Next
I had a family member die young from breast cancer. The information that might have changed that outcome may well have been there. The system just wasn’t built to act on it.
That’s one reason this work matters to me. A lot of what we call tragedy in health care is also, in part, a reflection of the lack of systems designed to help prevent those tragedies.
For most of my career, I’ve tested AI tools and often found them lacking or difficult to develop and deploy at scale. What’s different now is that the technology has become incredibly powerful and versatile. There’s still a lot of work ahead. But for the first time, it feels possible to widely implement the types of tools for supporting care that we could only dream of just a few years ago.
In the years ahead, patients and providers will expect better support: the right information faster, fewer missed opportunities, less paperwork, and more room for thoughtful care. I believe AI can help us get there.
