In this interview, Dr. Azizi A. Seixas, from the University of Miami Miller School of Medicine examines how fragmented data, workflows, incentives, and governance are slowing the adoption of artificial intelligence (AI) in healthcare. He highlights that the next five years are critical for the digital transformation of healthcare due to the convergence of mature technologies, clinician burnout and economic pressures. Dr. Seixas also shares the Virginia Mason Manufacturing System, a case study that healthcare leaders can look to as they consider how to implement and scale AI transformation in their organizations.
Dr. Seixas is an associate professor of psychiatry and behavioral sciences and director of the Media and Innovation Lab. In addition to these roles, he also serves as Deputy Director of the Center for Translational Sleep and Circadian Sciences and Interim Founding Chair of the Department of Informatics and Health Data Sciences.
–
Transcription:
How will fragmentation impact the adoption of AI in healthcare?
Dr. Seixas: I believe there are four places where fragmentation has the biggest impact on technology adoption. The first is really about data fragmentation, where patient data exists across multiple spaces, including electronic health records (EHRs), devices, imaging systems, wearables, and insurance claims. And when that data stops flowing, AI won’t be able to fully understand the patient’s story. This really limits everything from prediction to personalization, which I think is the true goal of the patient care experience.
The second area concerns workflow fragmentation. Many people in the health care system feel this, and indeed, so do patients. Because all too often, whether it’s adding dashboards, alerts, or portals, when you digitize broken processes without actually redesigning how care is actually delivered, you end up needing more. number of clicks but not necessarily more care. This is a major source of clinician frustration, and we see it often.
The third area concerns the segmentation of incentives. Many of these promising digital tools, from remote monitoring to AI triage to even asynchronous care, work best between visits, yet most reimbursements reward visits and documentation. Therefore, in-person synchronous experiences are often privileged and rewarding. I think that if the adjustment is off, the implementation may be delayed. [regardless of] Does the technology work?
And the last issue, which I think is even more important because it helps control other areas, is the fragmentation of governance. For example, decisions about AI, privacy, security, and operations are often siled across different departments across the health system. They do not actually interact with each other and do not provide a fully integrated operational system. This can lead to approval delays and mismatched standards. Frankly, there may be a great deal of mistrust on the part of clinicians.
So, the important point is that just because digital tools are immature doesn’t necessarily mean they fail. I think they fail because they are introduced into a fragmented system without consistency. And I think there is a serious problem there.
Why are the next five years key to digital transformation in healthcare? How can organizations capitalize on this moment?
Dr. Seixas: i really believe that [over] The next five years represent an important and decisive period for real healthcare transformation. One reason for this is that three forces are potentially converging, but unless we act now, they will unfortunately diverge. First, we are at a stage where technologies are becoming more mature, and general purpose and accelerated purpose technologies such as artificial intelligence, remote monitoring, and digital diagnostics with digital biomarkers are no longer experimental. It’s ready for production. Many claim that we are still in the pilot phase, but some of us believe that we are far from the pilot phase where people want actual implementation into the system. I think that’s where we are now.
Then the second force is about employee pressure, and one of the real keywords is burnout. Burnout remains a critical issue of great importance. It has been reported that over 40% of physicians report symptoms of burnout. The average amount of “pajama time” is approximately 108 minutes, requiring healthcare workers to spend additional time outside the clinic to perform essential tasks for patients. Healthcare won’t work if you keep adding administrative tasks. This is where I believe technology should ease the burden, not add to it.
And the third force, which could potentially converge but will become a problem if the right leadership is not found, is economic pressure. What we are seeing now is that inefficient care models are no longer sustainable as health system margins tighten and systems become more focused on efficiency, accountability, and value.
So how do organizations take advantage of this? This requires agency-level intervention, but the biggest mistake is buying the technology first. Instead, I believe leaders should choose two or three high-volume routes.
For example, consider my areas of focus: hypertension tracking, diabetes management, and post-discharge care. Any of these routes can be redesigned end-to-end. Let’s take high blood pressure as a specific example. Therefore, today usually a patient is examined and laboratory tests are ordered. Unfortunately, follow-up is often delayed, and deterioration in blood pressure may not be discovered until months after the patient visits the emergency room. …If we redesigned the route, it would look completely different. You can stratify risk during your visit. We can introduce home blood pressure monitoring. We do this in a lot of our work, building our own remote health monitoring solutions.
You can then let the AI filter your data. Clinicians can triage and see exceptions rather than noise. That’s a big thing to be careful about. Many people would say, “More devices means more data.” But my clinicians and friends often say, “Not all of that data is actionable.” [In the updated care pathway,] You can also have someone on the care team respond first, such as a nurse or navigator, after the AI filters out the noise. after that Some doctors may only escalate if necessary. The loop is then closed, perhaps with automated documentation and messaging to the patient. Then the visits are not yet digital, the intervals between visits are, and I think that’s the key.
So if institutions miss this moment, they won’t just be slow; They are tied to traditional workflows and, as a result, to vendor choices, forcing them to transform later in times of crisis rather than strategic situations.
What can the Virginia Mason Production System teach healthcare leaders about system-wide technology implementation?
Dr. Seixas: I think this lines up well with what I alluded to at the beginning. This is a powerful example of achieving this [transformation process] Yes, long before the AI hype.
Virginia Mason faced many of the same challenges that medicine faces today. This was a Seattle-area hospital that was plagued by safety concerns, inefficiencies, rising costs, and dissatisfied staff. Rather than start with technology, they adopted the principles of the Toyota Production System. That became the Virginia Mason Production System, which now has a research institute.
They focused on flow reliability and waste reduction, all really considered from the patient’s perspective. Because that’s how you become customer/patient-centric. They mapped the care process end-to-end and standardized their work. They eliminated rework and delays and created a culture of continuous improvement. Only once they achieved process clarity did they add additional technology to their deployment. I think this is why Virgin Mason’s framework became this national model. Not because of the tools themselves, but because of the operating system they built.
This is a lesson for healthcare, and it is a very profound one. What is operational excellence? Prerequisites For digital excellence. That’s the most important thing you can learn from the Virginia Mason production system model.
What vision should leaders have for technology implementation to successfully transform systems and practices?
Dr. Seixas: I think the vision that leaders and clinicians need to adopt is that technology is clinical infrastructure, not innovation theater. Technology must make care simpler, safer and more continuous. AI should support expert judgment, not replace it. And success should be measured by results and experience, not just implementation and dashboards.
For clarity, I usually use this acronym: FLOWS. Before you buy any technology, you need to fix your flow first. FL wants to reduce the burden on clinicians. O operates AI within the workflow. And the W is for working as a team, not in silos. And S represents the scale at which it works. If technology doesn’t improve flow, it’s not just a transformation.
What is one action healthcare leaders can take today to jump-start digital transformation?
Dr. Seixas: It is important to choose only one clear path. It’s important that people pay attention to what the process is. I collect what I think of as stakeholder maps of the ecosystem, take them into a room, and spend about 90 minutes mapping them. [workflows] from end to end. Here you can see where information is being lost, where decisions are stalled, and where more can be done. Next, test just one digital intervention that removes friction. Then measure your results after 30 days. These are sprints. Then you’ll see that transformation doesn’t start with purchasing software. It starts with clarifying care and the flow of care. Based on that, I think technology will eventually deliver on its promise.
