How AI-based technology can help improve hospital capacity management [PODCAST]

Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes

“Health system leaders can master this ‘chess game’ by anticipating the next several moves well in advance with the assistance of AI-based predictive analytic tools. This adoption will lead to dramatically better outcomes than a purely reactive response with no appreciation for how unit capacities are likely to unfold. In managing the next phase of the COVID-19 pandemic and beyond, it will become evermore critical for health systems to invest in transparent, data-driven solutions that will optimize capacity management issues.”

Pallabi Sanyal-Dey is an internal medicine physician.

She shares her story and discusses her KevinMD article, “How AI-based technology can help improve hospital capacity management.”

SUBSCRIBE TO THE PODCAST → https://kevinmd.com/podcast

RATE AND REVIEW → https://kevinmd.com/rate

FOLLOW ON INSTAGRAM → https://www.instagram.com/kevinphomd

FOLLOW ON TIKTOK → https://www.tiktok.com/@kevinphomd

GET CME FOR THIS EPISODE → https://earnc.me/RC2miq

Powered by CMEfy – a seamless way for busy clinician learners to discover Internet Point-of-Care Learning opportunities that reward AMA PRA Category 1 Credit(s)™. Learn more at about.cmefy.com/cme-info

Transcript

Kevin Pho: Welcome to the show, where we share the stories of the many who intersect with our health care system but are rarely heard from. My name is Kevin Pho, founder and editor of KevinMD. Rate and review the show at kevinmd.com/rate. Subscribe at kevinmd.com/podcast. And on the show we have Pallabi Sanyal-Dey. She’s an internal medicine physician. Her KevinMD article, which we’ll talk about today, is titled “How AI-based technology can help improve hospital capacity management.” Pallabi, thank you so much for joining me on the show.

Pallabi Sanyal-Dey: Absolutely. I’m so excited.

Kevin Pho: We’ll get into the article in a little bit, but first off, briefly share your story and journey to where you are today.

Pallabi Sanyal-Dey: I am now in the atypical clinical pathway, another way to continue being a physician. I started this journey a few years back. I’ve been part of the UCSF system for years and was fortunate enough to go through some Lean-based health care certification while part of the system, and I got to do a lot of different kaizens, process improvement, all that stuff. And I realized through that work that I actually really enjoy working on hospital operations, as challenging as it can be. I think the thing that really fulfilled me in that work is being able to meet folks and staff members in all different areas or sectors of the hospital, whether it’s housekeeping, transport, or the cafeteria, you name it, all the different people. Because there are so many people involved in really delivering good-quality care in a hospital, it really takes all of those folks, and understanding their flows and their processes.

So anyway, I really enjoyed that aspect of it, getting to connect with all these different folks outside of just my pure medical team of docs and nurses and whatnot. And with that, I started to realize that there are certain things in academics people strive for when they’re in that type of realm. I love teaching and working with medical students and residents, but outside of that, I found that with some of the operational process improvement work I would do, my hands were a bit tied, just because of the nature of being in certain types of systems, and I could only give so much effort to it. And then I also realized that in the long term, I wasn’t really focused on how many papers I can get published or what academic role I can achieve, etc. And then it led me to think, “What am I aiming for?” And I realized, after a lot of soul-searching and working with some coaching, even outside of UCSF, that opportunity lies outside of the traditional hospital setting.

So that is when I really embarked on a bit of a career search: What does a physician do if they’re into operations and wanting to explore tech and health care innovation? There’s no map for that. So what it led to was about 40-plus coffee talks, essentially. This was pre-pandemic, meeting with folks, even some who actually came from UCSF and then branched off to do their own startup companies or work in other realms, and other people within my network, just getting a sense of the different ways I could go in my career. I also worked a little bit on my LinkedIn profile, little things like that, so folks knew I was actually searching, which is also an important thing. It took a long time, about two years, I would say, of some dedicated time doing a lot of learning and conversations, cold calling, if you will, through emails, chatting with folks, and understanding their journeys, and then a recruiter actually reached out to me regarding an opportunity that lived in this hospital clinical operations space. Eventually, long story short, that is what landed me in my current job today, where I literally get to work on how you improve inpatient flow from outside of the system, leveraging technology and math, but using the experience I’ve had being on the front lines and actually working with teams to do some of this work.

Kevin Pho: So now, before getting into your article, “How AI-based technology can help improve hospital capacity management,” give us a little bit of a hospital workflow 101. Tell us about some of the biggest obstacles that really impede patient flow in the hospital, before getting into your article.

Pallabi Sanyal-Dey: So, some of the biggest things with hospital flow. At the end of the day, a good flow model is a model in which a hospital is constantly able to provide access to care for any patient who arrives, either through their ED, from the OR, or from an outside clinic. The moment they need to go into your hospital and receive inpatient care, it happens. There is no backup, there is no ED boarding, there is none of this nonsense, or you don’t need to get transferred to another facility. Now, there are a lot of elements to that throughput piece that make it challenging because of inefficiencies that are unfortunately inherent in the system. It is a little bit of the Swiss cheese model, where it’s not just on one party; it’s several elements that need to be in synergy to make those inefficiencies go away. But that is hard work, and we have imperfect systems, whether it’s the EMR, the manual processes a lot of our admin have to use in the hospital, or literally just staffing, which is the biggest issue right now across the nation, not having enough folks to even staff and really meet the demand inside hospitals.

As for the bottlenecks inside a hospital when it comes to flow, first off, obviously, as everyone knows, it often happens in the ED, where the gatekeepers are. In the worst-case scenario, the hospital, God forbid, goes on something called diversion, and they have to tell the ambulances to route to other hospitals, hopefully within their network, or other places. The second thing, which obviously happens more often than not, is patients do end up getting admitted, but they’re sitting in the ED, which is not an ideal environment for them to receive their inpatient care. Lots of studies have shown what those outcomes can be when they’re racking up minutes to hours to even sometimes days in the ED. And then when they finally get into the hospital, they’re not necessarily on the right floor for the particular service they have been admitted to, or by the time they get there, hospital day two or three has started.

And then in the care that they receive during their inpatient stay, there are inefficiencies there, like I said, usually due to staffing and some other processes and waste that are built in: when you can get that MRI, when you can get the EGD. It’s the weekend, so XYZ doesn’t happen, even though it’s a hospital and patients are coming in sick throughout the week, no matter what time it is. So a lot of that inpatient care often serves as a barrier even to getting things done, whether it’s a medical sort of intervention. And then there are a lot of nonmedical issues that again rely on other teams to be able to provide or deliver, whether it’s a PT consult, speech, OT, case management, or being able to communicate with the insurance provider.

And then it comes a little bit more towards the end of the stay, where again your ancillaries, such as case management, can do the best that they can do. But if your skilled nursing facility doesn’t have a bed, guess what? The patients can be waiting inside the hospital until they actually have that bed, and no matter how hard the hospital works, they can’t necessarily change that story. So anyway, that’s just a quick summary of some of the bottlenecks that happen, but hopefully that gives a little bit of that sense.

And the other thing is, we’re great in hospitals at reacting to fires, right? Whether it’s clinical, whether it’s a code, etc., we are wonderful at working miracles when fires occur. We are reactive. The thing that’s hard to do in hospitals, given the resources we have and the efficiency issues I talked about, is to be proactive. Also, the tools we use are quite outdated compared to other industries, even though we’re much more tech savvy than we used to be when I first started training. And because of even that aspect of it, we’re not used to actually utilizing or leveraging technology at its finest. We often don’t even have access to it, and often it serves as a barrier. So anyway, changing from the reactive to the proactive state, I think, is where that sweet spot will lie, where it can positively impact flow for us. And that is the arena in which I work, in which I help health care systems utilize that. How do you use a tool of that nature in a really impactful, positive way?

Kevin Pho: Yeah. So you talk more about that in your KevinMD article, “How AI-based technology can help improve hospital capacity management.” So talk a little bit about that article, for those who get a chance to read it.

Pallabi Sanyal-Dey: Sure. I switched careers a few years back, and this was my first foray into really being exposed to what AI even means, or machine learning, right? I dipped my toes partially in that whole realm, but it’s a very different thing to actually be working directly with folks who literally are creating a tool to help the system that I’ve always been a part of. What I learned along the way is that there is so much to be learned from the patterns that actually do exist in the way we provide care and the types of patients we see day to day in the hospital. And the data scientists, the best engineers within this area, are able to pick up on these patterns through the technology that they have built. In the company that I work in, our particular arena is that of machine learning models, where they utilize both. There’s always historical data, but even more relevant is the real-time current state, what’s going on in the hospital right now, and a whole host of variables that describe different units in the hospital, the patient population in the hospital, and also the providers on that floor, whether it’s nursing, physicians, ancillary, you name it.

And these models that the data scientists are able to create and design can actually help to reflect, “I can tell you with great accuracy what’s going to happen hours from now,” whether it’s from a bed balance standpoint or from an admissions and a discharge standpoint. I no longer have to constantly rely on the human, on the nurse, on the physician, to constantly showcase, “Hey, I think this patient’s going to be discharged X hours from now, X days from now,” because we are also running around doing so much other work that that is often the last thing that we end up doing. So what I came to appreciate is, when you do leverage technology like this, it actually can do a lot of that work for you to provide an accurate forecast. And then when you have the ability to really utilize solid forecasting that is actually reflecting what is going to happen in the future, folks like me and you working in the hospitals, and our staff, can trust what these machine learning models are forecasting. And then the beauty happens, because then you get to ask, “If I know XYZ is going to happen hours from now, what can I do right now to prevent that bottleneck from happening?”

And that is what I discuss, the specifics regarding some of those operational decisions you can actively, effectively make if you have technology like this in your hand that showcases hour by hour, even a day, two days, or three days out from now, what’s going to happen from a flow standpoint. You can actually start to etch away at a lot of those throughput time-based metrics that affect things that the C-suite and health care systems have to care about, like length of stay, things of that nature, which unfortunately, at the end of the day, is a thing that folks need to pay attention to. And the last thing you want, which happened during the pandemic, is that you can’t even do any research or surgeries, which are like the financial backbone of a lot of health care systems, or if you even could, you don’t have any beds, because more than half those beds are taken up by COVID patients. So how do you utilize and leverage the tech we have to make sure you will always have that bed available for when the OR patient needs it, from the PACU, etc., from the ED, from the clinics? And that is what I get into in the article, the different facets of that: what kind of operational decisions you can make, whether it is discharge-focused, even knowing which units in your hospital are going to be in the biggest demand later on in the day. Those are actually the discharges you should be focusing on the most to make sure they happen, versus telling everyone, “Work harder, work faster,” which is unfortunately a lot of the messaging that happens right now.

Kevin Pho: So give us a scenario, maybe a before and after, that really contrasts how AI-leveraged technology can improve workflow. It could be a hypothetical scenario, but give us a before-and-after picture of what your technology would change.

Pallabi Sanyal-Dey: One example is when staff and leaders come to meet in a hospital when it comes to flow. They do it in a variety of ways, whether it’s huddles, bed meetings, things of that nature. Now, traditionally, if you do not have our tool, what usually happens in those settings is people go through somewhat of a roll call situation, which is actually an improvement from the past, where they just call out numbers, such as, “These are the patients we have on this unit today. I think X patients are going to leave. I think Y patients are going to come in, maybe two from the ED, three from the OR,” things of that nature. So it’s a little bit of educated guessing and a little bit of knowing, because they spoke with the PACU, they spoke with the ED, and they know that these patients are coming in. And that’s where it ends. And then they go unit by unit, and the resources throughout the day, your hospital admin, also go unit by unit, checking in with everyone and then reacting to fires that happen where the flow occurs. It is a current situation that gets outdated very quickly, because numbers are constantly changing. And what is lacking in that scenario is a strategic focus, because there’s truly no forecasting being used of what is going to happen in the future. It’s all about just now and what already happened in the past.

So what happens when people use technology such as ours, such as iQueue for Inpatient Beds, is the conversation literally changes overnight, from the reactive rote, unit-by-unit checking off a list and answering pages, to, “I know we’re just going to talk about these areas of the hospital, because I know at 7 a.m. tomorrow, I know at 7 p.m. tomorrow, this is going to happen. They’re going to run out of beds. So what do we do right now to improve our bed availability, to improve that discharge outflow, to improve that throughput, so that they can actually meet their demand? Let’s talk about that,” versus everyone just roll-calling out, doing their check-ins of their numbers, and really no problem-solving in a proactive fashion happening in those meetings. It’s situational awareness that’s happening, but that’s the extent of it, and putting out fires. So we change it with a bit of a more proactive focus.

Kevin Pho: We’re talking to Pallabi Sanyal-Dey. She’s an internal medicine physician. We’re talking about her KevinMD article, titled “How AI-based technology can help improve hospital capacity management.” Pallabi, what are some of your take-home messages that you want to leave with the KevinMD audience?

Pallabi Sanyal-Dey: There is a lot of opportunity in using certain tools, certain things within tech, whether it’s machine learning, AI, whatever you call it. There is actually stuff out there that can help support us in our day-to-day, that actually will not be the barrier to our work but can actually elevate our flow work and make us do work in a smarter, more strategic fashion. We do not have to rely on our constant workarounds, our educated guessing, etc. We actually have tools that can help us in this arena, and if you choose the right tool, it can change the way you do hospital operations. At the end of the day, it literally can change and improve your ability to provide increased access to care in your hospitals. I want to shed some light and provide some hope in that arena.

And there are even some newer things we’re working on, such as being able to do natural language processing for whatever is being entered into the EMR, so we can signal to our staff what they can work on when they start the day, without having to wait for rounds, wait for calls, go into the chart, things of that nature. So the front-liners are not always carrying the burden, having to message, having to tell the story.

So my main message there is, please do explore these fields. There is some incredible innovation going on. Of course, I’m excited about what we’re doing with iQueue over at LeanTaaS, because we’ve been able to see some amazing results across a lot of health care systems. So I encourage folks to explore that, and to have that conversation, and to know that things can actually be better. It doesn’t have to be the way it is in its current state, and we don’t have the luxury of waiting either, given what we’ve seen in the pandemic. Flow has always been an issue; it just got spotlighted during the pandemic. This issue is here to stay. So get out there, contact folks like me, and explore what’s out there. There’s a lot to be used, and it can actually help make the staff experience, and also the patient experience, better.

Kevin Pho: Pallabi, thank you so much for sharing your time and insight, and thanks again for being on the show.

Pallabi Sanyal-Dey: Thank you.

Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes