The Role of AI in Outpatient vs. Inpatient Care
The integration of AI in healthcare is evolving, with unique challenges in both outpatient and inpatient settings. Outpatient care often focuses on improving efficiency through digital scribes and AI-driven documentation tools, helping doctors manage their time more effectively and reducing burnout. This can enable physicians to see more patients without sacrificing care quality. In contrast, the inpatient setting presents a different landscape, where the need for seamless data capture is crucial, especially when patients can’t communicate directly. AI solutions in hospitals must go beyond conversation-based documentation, utilizing sensors and advanced monitoring tools to ensure accurate patient data. The future of AI in healthcare will require tailored solutions for both settings, with an emphasis on improving patient outcomes and easing the burden on medical staff.
Listen to Narinder Singh and Eric Yablonka explore the evolving role of AI in outpatient and inpatient care, emphasizing the need for tailored solutions, seamless data integration, and innovative approaches to improve efficiency and patient outcomes in both settings.
Video Transcript
Narinder Singh:
I mean, conviction is a powerful thing that you believe in what you’re going to go through. I’m going to switch gears for a second and talk a little bit about outpatient inpatient. You ran both parts of this outpatient, I think we’re all familiar with the outpatient side. We go to our doctors all the time, think you become more familiar with the acute care side as you age or your parents’ age or the birth of your child or unexpected events we’ll talk about. Maybe you can talk a little bit about the differences from the IT perspective on how these two worlds operate.
Eric Yablonka:
Well, it is quite different yet so many acute care organizations own physician practices. Now you do end up leveraging many of the same technologies from the hospital to the physician office. So particularly in academic medical centers where I spent most of my career, that was not that unusual to have clinics or physician practices on the same platforms, but it is a very different world. On the outpatient side, there’s a lot of pressure on the practice teams to move very quickly. You’re 10 or 15 minute appointment or perhaps you’re seen by mid-level providers, PAs or NPS to relieve physicians of some of the burden. So you’re seeing a lot of different models on the outpatient side, but they all require electronic medical records. Many of the specialties require digital imaging, whether it’s orthopedics or pathology as an example, and they need that access from often the hospital systems.
And having just had a foot x-ray at one health system and going to another health system to get treatment, I was not delighted to have to bring along CD with me, which I recall doing 20 years ago. So image transfer is still something that seems like a mountain high to climb, but what’s really in that example, what was really problematic is that if I was in the same health system, it would’ve been no problem, but because I was in two different health systems, there was no interoperability. So on the outpatient side in particular, whether they’re independent practices or unaffiliated, that interoperability is a really big challenge. And so I think going forward as boomers like me continue to age and demand spikes, those technologies are going to be critical one to create more capacity. I don’t think the healthcare systems are going to be building a hundred new hospitals and 10,000 new physician offices and have the physicians to staff ’em. We’re going to have to look at tech tools to be able to handle both the influx on the acute care side and on the outpatient side. And it is all digitally enabled. So we have to find ways to leverage, otherwise people are going to have systems or they’re not. And we as patients, we as the civilians in healthcare are going to be very challenged to find care if the models don’t evolve.
Narinder Singh:
I like one thing you said, and I want to translate it into a couple of things around AI is that this kind on the outpatient side, it’s the visit and that interoperability is important. Not all sitting under one roof, right? You go to a cardiologist, you go to a podiatrist, you go to these different places are required to kind of take care of who you are. And the visit is like this unit, this atomic unit that floats around and we want to connect on the hospital side In some ways. We’ve done the integration by putting doctors from various specialties under one roof, so we solved it in a different way and it’s acute, so there’s a time bound aspect to that. And so there isn’t a visit in the same way, right? There may be visits, there may be different pieces, but care is much more than the time you’re talking to a doctor inside of the hospital. And I think that’s a profound point on the AI side specifically, especially on the outpatient side, generative AI has turned the world upside down and generative AI on the outpatient side has been a lot around scribes and documentations. There’s a dozen plus companies that have raised 10, 20, 30, a hundred, $200 million. There’s a gold rush around trying to solve this problem on the outpatient side. I’m curious on your thoughts on that and what of that you think translates over to the acute care hospital side and what is really distinct?
Eric Yablonka:
Well, I think sort of the generative AI around documentation is catching fire on the healthcare organization side. So both on the physician visit as well as in their practice. Let me back up a second. So really the question is, let me think about the question a little bit. So are you asking why don’t you reframe the question for me? Yeah,
Narinder Singh:
Alright, I’ll reframe and we’ll start from this. So Eric, we’ve seen the AI piece become profoundly everywhere all at once because of the generative AI component and particularly for healthcare on the outpatient side, scribes and helping doctors become more productive, whether it’s for burnout or do more visits in a day, there’s a dozen plus companies that have raised 10, 20, 50, a hundred, $200 million to solve this problem. What did that do you feel like translates to the inpatient side and where are those solutions going to be very distinct? Are we going to see this same act replay on the acute side exactly in just a longer period of time or is it going to be very different solution sets that you think matter for the hospital side?
Eric Yablonka:
Well again, we were just talking about the systems that need to interoperate with the EMR, and this is a great example where people are using digital scribes, ambient listening to generate notes that will take a lot of pressure off physicians and hopefully improve the patient experience because as we know, at least on the ambulatory side, the physicians are time constrained, they got to move and nobody wants them to do pajama time at night finishing all their notes and all their work. It’s just not healthy On the acute side, there’s also a lot of work around the revenue cycle using AI and of course that’s ripe for applications and use cases being so complex and so labor intensive. So that’s kind of where people have started in a large part, but there’s going to be a lot more opportunity for AI in the inpatient setting and there are a lot of issues that have to be addressed, whether it’s security standards, governance around AI and algorithms, A lot of that stuff is going to be addressed through congressional action at some point. People believe it’ll go very slowly, whether it’s because of the election or just how Washington runs, but we’ve heard a lot out of Washington about the concern around ai, but the healthcare organization’s going to adopt products that improve the clinical or operating performance of their organization and it’s just a matter of time. It’s the way it always has been. With any of the technology, digital imaging, everything was on film until it wasn’t
And it wasn’t because it was a better way of taking care of patients and more cost effective.
Narinder Singh:
I think in the digital imaging side, you really reduce the friction of participation and to hear, I think the generative AI on scribes does reduce the friction on the outpatient side. You’re in a different situation and there the voice, the conversation really is the mechanism for that. On the inpatient side, there’s a lot more than the conversation, right? Many times the patient can’t even talk. My mom’s case, she was trached the entire 12 weeks. There was no conversation to document was not the essence of peace. It was this ability to kind of sense what was happening without requiring somebody to articulate that or engage in a conversation to get to it. And so I think there’s analogies, but there’s obviously a profound difference when your conversation is such a small percentage of care just if you look simply by how many minutes a day doctors and nurses are engaging with the patients. I want to
Eric Yablonka:
Norin if I can interrupt. Of course physicians and nurses spend hours and hours and hours on every shift documenting things and that manual process of documenting whether it’s operative notes or history and physicals, those kinds of things, they can be helped as well. The acute care documentation burden is profound and people think EMR has brought that on and they’re only built that way for billing requirements. At least the medical staff usually believes that and these kinds of tools can help with that as well. On the inpatient side,
Narinder Singh:
A hundred percent agree and I think but it’s different, right? And the outpatient side, you and I are having a conversation about your ankle or your foot on the inpatient side. I just want to see what I’m doing and document it for me. I don’t want to have to have a conversation and talk to myself and in fact sometimes it’s not even feasible. I just want to know what’s happening with the patient when I’m not there or what I’m doing with the patient and hit yes. I think the goal of getting people out of entering and typing into the system of record is shared across these things, how it manifests itself, is it language driven or is it sensing driven with video and sensors? I think that’s where there’s a pretty profound difference between the underlying technologies that will make those occur. I want to shift gears for a second.
I want to shift into what we think about as we think about this flexibility, we think about the AI components and how we can help be the best at that. But I want to shift to the last part around alignment and we think about this from a couple different things.