Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes
We sit down with Maria Iliakova, a bariatric and general surgeon, to explore the exciting ways AI tools like ChatGPT are reshaping clinical practice. Maria shares her experiences using AI to enhance hospital staffing, analyze patient data, and predict patient flow in real time. We discuss how AI can help address health care staffing shortages, improve the efficiency of call schedules, and provide more personalized care.
Maria Iliakova is a bariatric and general surgeon.
She discusses the KevinMD article, “Can AI truly improve hospital staffing?”
Our presenting sponsor is DAX Copilot by Microsoft.
DAX Copilot, by Microsoft, is your AI assistant for automated clinical documentation and workflows. DAX Copilot allows physicians to do more with less and turn their words into a powerful productivity tool. DAX Copilot automates clinical documentation—making it available in the EHR within minutes—and clinical workflows, including referral letters, after-visit summaries, style and formatting customizations, and more.
70 percent of physicians who use DAX Copilot say it improves their work-life balance while reducing feelings of burnout and fatigue. Patients love it too! 93 percent of patients say their physician is more personable and conversational, and 75 percent of physicians say it improves patient experiences.
Discover AI-powered solutions for clinical documentation and workflows. Click here to see a 12-minute DAX Copilot demo.
VISIT SPONSOR → https://aka.ms/kevinmd
SUBSCRIBE TO THE PODCAST → https://kevinmd.com/podcast
RECOMMENDED BY KEVINMD → https://kevinmd.com/recommended
GET CME FOR THIS EPISODE → https://kevinmd.com/cme
I’m partnering with Learner+ to offer clinicians access to an AI-powered reflective portfolio that rewards CME/CE credits from meaningful reflections. Find out more: https://kevinmd.com/learnerplus
Transcript
Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast. Today we welcome back Maria Iliakova. She’s a bariatric and general surgeon. Today’s KevinMD article is “Can AI truly improve hospital staffing?” Maria, welcome back to the show.
Maria Iliakova: Thank you so much, Kevin. It’s really nice to be with you.
Kevin Pho: So we were talking offline, and I know you’ve been on the show before, and you’re a bariatric and general surgeon, but now your career has taken a little bit of an AI tangent. So tell us what that is.
Maria Iliakova: Yeah, I think a lot of people’s careers, honestly, have been taking a little bit of an AI tangent. It’s definitely the buzzword that’s around everywhere right now. It’s big and investing right now; a lot of companies are getting started around AI. So, funny enough, I actually worked on AI projects a long time ago, when I was doing a master’s in bioinformatics. I actually did some protein structure calculations and modeling, not the cool kind, but the kind that you do with Python, ages ago, like more than a decade ago. And it’s great to see these kinds of tools take a quantum leap, honestly, since that time.
So what we’re able to do with AI tools now is just leagues beyond what we were able to do just five, ten years ago. And some of that is actually some interesting factors driving that. One is processing power has changed radically since that time, too. So imagine what cell phones used to look like even ten years ago, or computers used to look like ten years ago, and now you can essentially carry an entire library of songs, or videos even, in your phone, which isn’t something we could even ten years ago. So the processing power, computing power, has changed radically.
And then also, concepts like big data, for instance, have matured a lot in the past ten years. We have databases of everything from patient information, for instance, to molecules and the ways that they interact, to even lots of data that Google, for instance, has been collecting on all of us and how we use our devices and technology. Though they do it for advertising purposes, it can obviously be deployed for other purposes, too. So yes, my career has yet again taken a bit of a detour into AI. Yes.
Real physician voices, twice a week
Free, and one click to unsubscribe.
Kevin Pho: All right, so we’re going to talk about one of those AI-related articles. It’s titled “Can AI truly improve hospital staffing?” So tell us why you wrote this article in the first place, and then, for those who didn’t get a chance to read it, tell us about the article itself.
Maria Iliakova: Sure. I’d be surprised if many people actually read it, but in short, basically, I think in health care, we know of AI, we’ve heard of AI, but we don’t necessarily think we’re using a lot of AI, necessarily, and we don’t really know how it’s going to be applied to what we do or how it’s going to change the work that we do. So this article is addressing that, but specifically asking the question of, you know, what about one really, really important question and really important topic, which is hospital staffing?
And hospital staffing is something actually that, you’re wondering, probably, why on earth is a surgeon talking about hospital staffing? Well, let me tell you, it is the bane of our existence when we don’t have staff in the OR, for instance, because it delays our cases and delays our patients. So over 90 percent of hospitals at this point are actually using locum physician staffing. So that’s not just temporary nursing staffing; that’s temporary doctor staffing. And that’s 90, 90 percent, not 19, 90, which I think is pretty shocking. I didn’t know that until I started looking into it.
And I’ve had the privilege of actually working as a locum surgeon in hospitals in Iowa, in Missouri, in Florida, and a couple of other states, and have really started to understand that this is a crisis, especially for our rural and community hospitals, but it also reaches into cities and academic institutions as well. So, you know, it really crossed my mind: What if we could use some AI tools to make hospital staffing a little bit better? We can’t just create new doctors out of nowhere. Creating more residency positions is literally an act of Congress. So you can’t just materialize people out of nowhere, but what if we could use the people we have? And that’s exactly where things like AI can help us, is to use the resources we have in a way that maybe people haven’t thought of yet or people aren’t as good at doing.
This is combined with, you know: The hospital where I was working in Iowa City, Mercy Iowa City, went bankrupt last year and became part of the bigger university institutions basically next door there. And one of the biggest factors driving that bankruptcy was staffing and the inability for this small independent community hospital that had been standing for almost 150 years to actually staff physicians. So, you know, it’s a lot of different opportunities, but at the same time, I figured it would be really cool to ask this question.
In this article, I specifically talk about something that I was learning in one of my MBA classes. So on top of all this, I also started an MBA, a full-time tech MBA at NYU, in May this year, and we actually do coding, Python, which is a programming language, and we use it in something called Google Colab, which is a notebook that you can use to deploy code. And I was blown away. Like I told you, Kevin, I have done programming and things in the past, but the ability to have this powerful of a tool, and not just to be able to start programming and coding right away, but the AI tool that was built into this notebook (it’s called a Jupyter notebook; it’s created by Google Colab) literally was writing code for me as I was commenting on it.
And the kind of insights you can get from that, the kind of understanding of data, the kind of images you can generate with that, it blew me away, because it’s not just ChatGPT and getting some random answers to things. You’re actually generating the code itself, so you understand exactly how the results you’re getting are being produced. So anyway, I really wanted to share my experience, because as a doctor, I have not been coding for the past ten years or so. So even someone like me could generate this in a few minutes using some powerful tools that involved AI. So that’s what this is all about.
Kevin Pho: So tell us about some of these AI tools. And how can they address some of the hospital staffing issues that you brought up earlier?
Maria Iliakova: Yeah, absolutely. So I think we’ve all really felt the crunch, even if you’re in a place that happens to be really well staffed. We all were affected by COVID, and we all saw people leaving the workforce and some not reentering, some people not reentering full time. And then even people who are coming into the workforce are also much more cautious about signing up for full-time work or going to certain locations where there may be greater instability than others. So it’s a really big deal.
And then when you start asking questions as to, like, how can we fix these problems? We can’t just generate more doctors. What AI is really, really good at are things like sorting, recommendation, prediction, basically analyzing lots of data and coming up with trends, coming up with insights on that data.
One of the biggest hurdles that I encountered as a locum physician in staffing was actually credentialing, which you wouldn’t necessarily think. That is not a sexy topic. Nobody’s, like, awake at night thinking about how cool credentialing is, maybe except me at this point, but it’s a huge hurdle. So it can take months. I think every single physician has encountered a credentialing problem that they had, where something just didn’t come through on time or they were left waiting for a week, weeks, or months before they could start a job, especially if you’re transferring states, because we do state licensing in this country for our doctors.
And I mean, the last time that we actually credentialed somebody for a work assignment, it took thirty-seven emails, seven and a half hours of work for five different people. And that was between a really motivated hospital staff, a really motivated surgeon, and a process that was trying to move as fast as it could, and it still took that long. So I think every single doctor who listens to this will relate to how annoying and time-consuming credentialing can be, especially when you’re trying to get a new job and just really excited to get started.
At the same time, from the hospital side, it’s a huge annoyance, because it’s losing them the ability to take care of patients, losing them money, losing them credibility. And if you don’t have a doctor in a program, let’s say a GI doctor for an endoscopy lab, you don’t have nurses either. You don’t have techs either. If your doctor isn’t working, the whole team isn’t working, oftentimes, whether that’s in a clinical setting or an inpatient setting. So it’s really a problem with reverberating effects.
So I started a company, initially called Search, but we’ve actually rebranded. Doc Dogo is the name of the company. What we do is we basically take credentialing documents from doctors, and we have them upload them just as a document dump, basically the simplest thing that you can do, and we actually deploy AI on the back end in order to analyze the documents and fill out, basically, our digital forms for you, to create a profile, to create a CV, and to create something that can be shared with employers, with hospitals for jobs, with insurance companies, with electronic health record companies, and others. And it’s something that can help you update your credentials, keep track of your credentials, renew them when needed.
It’s actually kind of shocking to me that something like this doesn’t already exist, but, you know, the best thing out there that I’ve seen is FCVS, and that’s not comprehensive, and it doesn’t really help track things, and it’s still kind of difficult to use, honestly. So we deploy AI to be able to do a lot of that manual work that doctors otherwise have to spend dozens of emails doing or filling out forms by hand, and AI just makes that process a lot faster and actually a lot more accurate, too.
Kevin Pho: So just give us a sense of that piecemeal approach that physicians have to use for credentialing. You mentioned that they have to have, like, dozens of emails just to get documents. Just give us a sense of how many parties are involved, how many documents a physician has to give out and look for and hunt down. Just give us a sense of that complexity.
Maria Iliakova: So it’s a great question. And kind of the most difficult part of this is that even though we’re doing this over and over again, it’s pretty much the exact same documents that every hospital is requesting every time. So it’s really, really a process that’s ready to be standardized, and it’s ready to be made the same between different hospitals. But what you’re trying to find for yourself are things like your degrees, your diplomas, your certificates, your credentials for your residency program, fellowship, your state licenses, your DEA, your CDS, your ATLS, BLS, ACLS. Your CV is part of it. Even your medical malpractice coverage is a part of the records that are requested, and even something like immunizations. And Kevin, I was born in Russia, so half of my immunization records from being a child are in Russian. So let me tell you how easy that is for hospital administrators to read and be able to use as part of my credentials.
So it’s really a field in which, you know, you’re trying to get all these documents, and some of them are handwritten documents. Some of them don’t exist online. You just have, you know, this thing that you’ve scanned a million times, and it barely looks like English anymore, even if it started out as English. And it’s something that’s ready to become something that doesn’t require us to spend hours or days hunting down documents and then emailing these documents, many of which are actually pretty private, like your DEA information or, you know, your immunizations, your medical records. We’re literally emailing these back and forth with administrators.
And on average, this process takes three to six months. It costs over $1,000 to collect and verify all of these documents per time that you do it. And most of these kinds of credentials, like your state license, BLS, ACLS, that kind of thing, they require renewal every two to three years, and you’re spending quite a bit of money and quite a bit of time making sure that all of that happens. So I really don’t know a doctor, if you talk to them about credentialing, that hasn’t gone through a pretty dramatically negative experience doing this.
Kevin Pho: So after using this AI approach, how much time does it cut off the typical credentialing process?
Maria Iliakova: Yeah, so once a doctor uploads their records, and we actually are involved in helping them find their records, and in some cases, we’re actually able to integrate with other databases online, like NPI databases and others, to be able to pull those records automatically without the doctor even having to upload those records. Kind of cool. It becomes a minutes-to-hours process rather than a months-long process. And that is a pretty interesting proposition, not just to the doctor, because it’s convenient and makes life a lot easier, but really to hospitals, because now hospitals can hire and verify credentials in a way that simply didn’t exist before.
And there are some incredible administrators that we’ve worked with to go through these kinds of credentialing processes. Every single person that I’ve worked with at every hospital that I’ve had to credential as a locum physician, I think, is a saint, because they go through this and basically, by hand, document by document, make sure that everything on that checklist is there and is appropriate. But that’s a really great system to automate and not have to have a person do.
On top of that, the administrators that are typically helping with credentialing typically have other jobs, too. So it’s not like they’d be sitting on their hands if this was taken off their plate. I don’t know a single person whose only job is credentialing for doctors or making sure that all those records are right. It’s usually somebody who has a lot of administrative hats, and sometimes I’ve even interacted with people like CEOs of hospitals or CMOs of hospitals or even CFOs of hospitals in this process, because it’s such an important process, but such a really difficult one sometimes, in the way that we do it now. So it really frees up a lot of people to be able to actually do more meaningful work. And also, even just reducing the burden off of doctors, to not have to spend hours to days hunting things down, I think would be really wonderful in my life, personally. It’s just a personal product I really wanted, too.
Kevin Pho: So just to be clear, it cuts down that months-long process to something that now takes hours.
Maria Iliakova: Yeah, minutes to hours, depending on how readable the documents are. The cool part of the process, too, is we’re actually testing with a couple of hospitals on the hospital side as well. So they’re able to upload their documents, and their documents get digitized in exactly the same way. So think of, like, your onboarding documents being able to be autocompleted with the records that we already have for the doctor you’re hiring. So again, that process becomes, instead of someone having to send a bunch of emails and manually fill out forms, whether it’s the administrator, the doctor, or both, it now literally becomes a digital process that happens automatically once a doctor and hospital are matched in the system.
So yeah, think of it, it’s also the same way. I mean, it really could be applied to a lot of things, any kind of application process for school. I would have loved to have done this for my MBA program, where I could just click a button and my entire CV gets uploaded, and I don’t have to sit there manually typing it out and filling out a form. But this is exactly what AI processes and computerized digital processes are made for, is to get rid of that busywork, to get rid of what we call in our industry scut work, right? Nobody wants to do that.
Kevin Pho: So in your article, you touch upon other ways AI can improve that hospital staffing issue nationwide. So what are some other ways that you talk about?
Maria Iliakova: Oh, yeah, absolutely. So some really interesting things that we’re hoping to come out of this work that we’re doing, this project, and that many others are also working on, is actually figuring out: What kind of staffing do we need? When do we need the staffing? Who do we need? How many nurses do we need on a floor? How many patients are we anticipating? How many surgeries are we actually performing on what days? Are there peaks? Are there trends? Do we need to be staffing two surgeons at this time? Do we need to be staffing one? Do we need to be staffing a specific kind of doctor versus a specific kind of nurse for different things? Again, that’s a really great field to use AI recognition, recommendation, and prediction tools to be able to figure out.
Some other interesting things that I think are great: You can even place not just time but location data into this and figure out, for instance, if a hospital has more than one location, or not even more than one location, but they transfer to other hospitals or they accept transfers from other hospitals. Those are also trends that you can evaluate over time, and you can come up with ideas as to what you can expect in terms of transfer volume to or from your hospital, where you need to deploy what resources to address the patient volume that you’re anticipating.
So again, it just becomes a really powerful way of helping hospitals schedule the appropriate workers for the work that needs to be done, because you and I both know that doctors and nurses sitting around doing nothing does not help anyone, and most doctors and nurses and people in hospitals like to be busy if they are working. And it’s not like you can do other work. It’s not like, “Oh, we can do a podcast while we’re on call.” It’s not really. Or, you know, you can’t go garden, or you can’t go get groceries or something else. If you’re in the hospital, you’re in the hospital. So we might as well be using the staff that we have in the best way possible. And also, it doesn’t make sense for hospitals to be paying staff to just sit around if that’s the case. So it’s really critical to match volume and demand with supply, and that is another application that AI can really help us with.
Some other things that kind of drill into the details of that are dynamic staffing, so, for instance, being able to have appropriate backups available or being able to shift people or flex people between different units, specifically for nurses or techs or nursing assistants, MAs. So I mean, really, the applications are pretty endless in how we can improve staffing from where we are now.
One other huge thing that I haven’t mentioned, but I think is an enormous opportunity for us, is to improve telehealth staffing. So for instance, neuro has been around in the U.S. for about a decade, and it used to be that less than 20 percent of the country had access to a neurologist to diagnose and manage a stroke within 30 minutes, which is the national standard. Now over 70 percent of the country does, because telestroke has become the standard of care in many locations. And there’s a process of credentialing (shocker, I know, I’m going back to that topic) of being able to allow neurologists specifically to be proxy credentialed, meaning if they’re affiliated with an institution, say, NYU, for example, if there’s a real hospital that contracts with NYU to recognize their credentials, they actually don’t have to go through the credentialing process all over again. Their credential at NYU serves as a proxy for them to be able to virtually see patients at another hospital and assess them for stroke and to be able to deliver that timely care.
As a result, morbidity, mortality have dropped where this is deployed, satisfaction has certainly dropped, cost has dropped, and now you have real-time access to a neurologist to be able to diagnose and treat a stroke for a patient, which is lifesaving care in many cases. So being able to apply these kinds of AI tools and credentialing, and being able to even allow hospitals to adopt more telehealth, even in the procedural space, which is completely, currently unexplored but starting to become possible because of robotic consoles. Yeah, there really are endless applications.
Kevin Pho: We’re talking to Maria Iliakova. She’s a bariatric and general surgeon. Today’s KevinMD article is “Can AI truly improve hospital staffing?” Maria, as always, we’ll end with some of your take-home messages to the KevinMD audience.
Maria Iliakova: Yeah, so I think one of the biggest ones is, when we’re deploying AI tools, there’s a lot of opportunity, but there can be a lot of cost as well. So something we didn’t really touch on, but I do think is an important takeaway, is that AI is a tool, and every tool, it matters how you deploy it. If you were to use a scalpel to try to fix someone’s diabetes, that probably is not going to work. So similarly, if you try to use AI in an inappropriate way, or for something it’s not made for, without thinking about how to deploy it, it not only may not get the job done, it might cause harm.
So it’s really important to engage people who know how to use AI, who are thinking about the social, ethical, and even privacy concerns involved in using AI, before it is something that becomes systematically used, or used in something as high value as health care. So that’s a really big takeaway we didn’t really talk about, but I think it’s really important to mention.
And then two, I think, starting to think about what AI can do for us to make us better. AI at this point, or at any point in the near future, will not replace doctors, and that’s great news in a way, I hope, because what doctors do that cannot be easily replicated, or replicated at all, by any technological tool so far is to validate results and to take responsibility for those results. So it’s liability, basically. For better or worse, doctors are uniquely capable of assessing and making a final decision for a patient’s care, and the liability associated with that, and there is no tech tool, AI or otherwise, that replaces that extremely valuable asset of a physician and what we’re trained to do.
So thinking of AI, and this is kind of the third takeaway: Two is, basically, AI is not going to take your job; it’s the doctor. But three is, AI can really, really help you do your job better, because if you know what to anticipate in terms of your patient volume, if you know that you’re going to have the staff that you need, if you know that you can basically sign up for a job and go rather than wait three and a half months or four months or five months to get your credentials ready, if you know how to take care of your patients better because a lot of the scut work is being done for you, these tools can be very powerful in how we use the resources we have now, use the number of doctors we have now to actually take care of this nation’s needs and nation’s hospitals.
Kevin Pho: Maria, thank you so much for sharing your perspective and insight. Thanks again for coming back on the show.
Maria Iliakova: Thank you so much, Kevin. It was such a joy to be with you.
Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes
