HIMSS 2024 Recap: Generative AI Uses, Workforce Management

Rich Birhanzel:

[Intro clip] Patient access is the global crisis that we have in healthcare today. We don't have enough humans that are trained or could be trained to serve the healthcare needs. We have to embrace technology.

READ MORE: Diabetes Tools Do Not Deliver, Epic Adds Ambient Listening

Kelsey Waddill:

Welcome to Season 2 of Industry Perspectives, coming to you from HIMSS 2024 in Orlando, Florida. I'm Kelsey Waddill, Multimedia Manager and Managing Editor at Xtelligent Healthcare Media. And if you didn't get to go to HIMSS this year, first of all, I'm sorry, it was awesome. And second, this episode is for you.

Today, we'll get to hear all the highlights from Rich Birhanzel, senior managing director and global health lead at Accenture, as he and I sit down to discuss some of the main themes of the conference. Rich also offered some of his own insights on workforce management and generative AI. He covered it all very succinctly, and I think you'll really appreciate his perspective. So without further ado, here's what he had to say.

Thank you, Rich, for coming on to Healthcare Strategies. It's a pleasure to get to sit down with you in person. We don't get this kind of opportunity very often. So I just want to start out by talking about, I'm sure you've been having a lot of really interesting conversations here at HIMSS. What are some of the ones that stand out to you? What are the topics that are swirling around here that you think are the most important for people to know about?

Rich Birhanzel:

READ MORE: How Healthcare Technology Can Improve, Worsen Provider Burnout

So the two things that we've noted--and it's kind of what we expected--AI is everywhere. Everybody's talking about artificial intelligence and what the possibilities are, what the worries are, what progress has been made. So there's a whole lot of conversation about that. And then the second thing is the labor shortage that exists on the clinician side, the challenges that the health systems and hospitals are having in being able to serve as many patients as they were able to do before.

Kelsey Waddill:

And I'm sure those things can get tied together sometimes too because a lot of people are turning to AI as a sort of relief for workforce management right now.

Let's talk about the AI part first. Obviously, we've had a boom in generative AI use this year. Have there been any really interesting kind of use cases or implementations of generative AI that you've seen that you thought [were] maybe going to pick up steam this year?

Rich Birhanzel:

READ MORE: How Artificial Intelligence, Machine Learning Impact Care Coordination

So the progress in healthcare around gen AI--and we're seeing it here at the conferences as demonstrations--has been in maybe four areas.

Contact center is the first because it's very much focused on language and generative AI is particularly useful when you're talking about language. And so the ability to better enable call center agents and create some efficiencies in some of the work that they do is sort of the first place. And we're seeing quite a bit of maturity in the call center space with generative AI.

The second space is the delivery of technology. So, thinking about how you can use some of the generative AI tools to create efficiency in the creation of code, the testing of code, and whatever projects that health companies are pursuing. So this has come fairly quickly over the last couple of quarters, where we're seeing a maturity in the ability to change the way we think about technology delivery.

A third area would be in learning. So again, learning is typically a highly language-focused activity. And so we're seeing quite a bit of progress in the use of generative AI to reinvent the way that we're doing learning. And then also learning is a focused area on teaching people how to get more familiarized with generative AI and be able to use it. Whether it's a call center agent, a claims processor, a physician, a front desk person, whatever it may be, there's a certain amount of understanding that needs to exist.

And then the fourth area is the one that people get the most excited about and you see a lot of this energy over on the show floor, which is around the use of ambient listening technology to listen to a patient and physician, or a patient and nurse conversation and automatically create a patient summary to reduce the amount of administrative burden that clinicians have.

Right now, we see about 20 percent to 40 percent of a clinician's time being focused on documentation. And what these tools have the promise to deliver is to remove that kind of burden. Sometimes that's done while they're talking to the patient, so that can become more efficient. Oftentimes it's done after their workday is done, so then we can give them the opportunity to reset, refresh and get ready for the next day. So a whole lot of energy around these emerging technologies, and these are in the field now. We're seeing experience with it now, and it has the promise to address some of the clinician overload that we see and lack of clinical capacity that is a problem for all of us.

Kelsey Waddill:

Yeah, absolutely. I want to get to that in a second. I was curious, one of the big question marks with AI right now and more broadly in the healthcare system--especially with the Change Healthcare attack--is privacy and security around AI. I'm wondering if you've been having any conversations with people about that and what have been the prevailing thoughts about how to secure these newer technologies?

Rich Birhanzel:

So there's a significant amount of conversation about responsible use. And one of the elements of responsible use among others is the consideration around privacy and security. The point of view we have and what we are seeing people talk about is that we need to think about responsible use before we go deploy these use cases into the field. It has a lot to do with what's the data foundation that the generative AI or AI is based upon. If it's something that's controlled and known inside of a health system, then it's sort of got a certain confidence level because it's data they already have.

When you start to extend that into other data sources, then it becomes a little bit more difficult. And being able to understand what exposure is being created when we deploy these models is essential. And the key part of that is it has to be done early before it gets deployed. And we think about that as part of the responsible use space.

There's other things in there like making sure we don't see bias in the data or bias in the recommendations that are coming back, that we are mindful of health equity as we deploy these tools. Those are examples of things, but there's been a considerable amount of conversation about that. And I think the other consideration here too is the use of generative AI, or AI, to also improve information security and use it as a tool to enhance and improve the way that health organizations protect their data.

Kelsey Waddill:

Makes sense. I'm sure we could dive into that a lot more, but I did want to make sure that we get to the workforce management piece.

Rich Birhanzel:

Sure.

Kelsey Waddill:

And so, obviously, last year we had the Kaiser Permanente strike, partly over this very issue. We've seen a lot of outcry among the provider population about the workload that they're experiencing right now, the administrative burden that they're experiencing. Where is the focus right now in terms of how to address these outcries?

Rich Birhanzel:

It's a great point, and it is a top topic. Patient access is the global crisis that we have in healthcare today. It's not just a US issue, it's an issue around the world. And the truth is we don't have enough humans that are trained or could be trained to serve the healthcare needs that we have around the world. So we have to embrace technology to create efficiency and make better use of the talent that we do have.

The areas of focus right now are about enablement. They're about taking administrative tasks that a clinician, nurse, physician, whatever role on the care team might play and take that role, think about the administrative task, like the documentation of a patient summary. That's a good example. Or a discharge plan as another example. Generative AI and particularly connected ambient listening can do that now, the tech is here. But to get it right, we have to harmonize that tech with the work of the clinician.

So when we take those tasks away, you might have a physician--let's say that part of their normal course of engaging with a patient would be to take handwritten notes or they might prefer or choose to type right into an EMR. That's a cognitive process that has been matured over time. And sometimes they do that not just to document, sometimes maybe they're doing that to catch up to their thoughts and think about what they're going to say next to the patient. But when we take that away and it's just a straight conversation that's being listened to by a computer, and then it's nice that it's going to create a patient summary later, we're changing their cognitive process.

So the energy that needs to happen now to really bring this out at scale is to harmonize that technology with the humans that are going to benefit from it. And that work is in its very early stages. There's much more to be done there to think about how we connect that human experience and improve their work so that we get the end goal here, which is to reduce the burnout, to create some capacity so that the clinicians that we do have can accomplish more and have better experiences in the work that they do.

Kelsey Waddill:

Yeah, can you talk a little bit more about that harmonization process? What is involved? I mean, I imagine that involves a lot of conversation between providers and the innovators who are creating these?

Rich Birhanzel:

Yeah, so it's early stages right now, but the idea with this is to go look at... Let's just take the clinical visit as an example, a typical clinical visit. What does the process usually look like? A nurse might start by taking a few measurements, asking a few questions, some diagnostic, and enter that into the EMR. That might be a typical thing to do. And then after some time waiting for the doctor to come in, if that's what the next step is, then the doctor comes in, might verify some of that information, then start to have a more thorough diagnostic conversation. "What's going on? How are you feeling?" And so forth.

And what we're talking about here is taking that process and then re-imagining that process. Imagine a world where the nurse doesn't have to ask all those questions, or particularly doesn't have to document all those questions. We might not even need to do that process step at all. But if we do need to do it, it might just be the nurse talking to the person and we capture that information and automatically update the clinical record. And then the physician comes in the room, and again, they don't need to go back to the EMR, they don't need to go look at handwritten notes or whatever their source might be. They can just have the conversation. They'll know what the information is, they'll have some sort of display and the listening capability will pay attention to the conversation. It will store what's being learned back into the clinical record.

And if there's an order given of, "hey, we need to go prescribe this," or "we need to go do a certain test," that also automatically feeds in the background to create the order, to create whatever needs to come next. So when we do that, what we're really talking about is re-examining both the process, but also the roles of the people in the process--particularly the clinician, but in some cases, in some ways, even the patient. Because it'll feel very different to a patient when you don't have this back and forth of, "I'm going to tell you something, you're going to go write it down." It's just a conversation. We think it will feel better in the long run, but it might feel awkward initially. So we'll have to get through that learning curve.

Kelsey Waddill:

Yeah, that's super interesting because almost like a blend of workflow change, but also almost like psychology and philosophy, too.

Rich Birhanzel:

Yes. Very much behavioral science in this, because you're really changing a cognitive process. And what we want to do is really figure out what is that way of leveraging the tech that works best so that the physician and the nurse can do what they're really great at, and the patient can have a good experience, and we remove the things that can be handled easily by technology.

Kelsey Waddill:

To free up the workforce so that they have more time to actually see patients.

Rich Birhanzel:

Yeah.

Kelsey Waddill:

Yeah, amazing. Well, let's hope that this stuff kind of gets off the ground and we can get that sort of harmonization down quickly.

Rich Birhanzel:

Yeah.

Kelsey Waddill:

Well, thank you Rich so much for coming on to Healthcare Strategies, it's been a pleasure to talk with you.

Rich Birhanzel:

Yeah, thank you. Appreciate it.

Kelsey Waddill:

Listeners, thank you for tuning in to Industry Perspectives. Subscribe to our Healthcare Strategies YouTube channel to watch interviews, get sneak peeks of upcoming content and connect with the Healthcare Strategies community. And as always, if you like what you heard, then subscribe to Healthcare Strategies on Spotify, Apple and other podcast platforms to get new episodes as they air. We have more episodes of Industry Perspectives on the way, so stay tuned.

• Tagged
• Artificial Intelligence
• Cybersecurity
• Healthcare Strategies
• Workforce Management