AI—Advantage or Danger in Healthcare?

Welcome. You're listening to Gravity Healthcare Hacks with your host, Melissa Brown, Chief Operating Officer from Gravity Healthcare Consulting and Self-Professed Healthcare Nerd. Monthly, we will provide industry expertise and tips to help keep your feet firmly on the ground in the world of healthcare. Episode 54, AI, Advantage, or Danger in Healthcare. Hello everyone, welcome to our podcast today. This topic is something that I am personally very passionate about and I really can't wait to hear what our guest, Richard Kedziura, co-founder and COO of Ascendist Solutions, has to say about this topic. Welcome, RJ. Thank you for having me. I'm going forward to the conversation. Great. I'd love to answer that question right away, but we'll let that come out during our conversation. Excellent. Well, tell our listeners a little bit more about you just to kind of level set and give them an understanding of why you're an subject matter expert in this and why you're somebody that we should listen to in the topic. So how did you end up starting a stand-up? Yeah, I got my start in grade school with technology. I was fortunate. I was so way back in the 80s, which is crazy that I can think about that, but I had a personal computer at home and I played lots of games. Don't get me wrong. I definitely enjoyed playing the computer games, but I was very fascinated by what I could do with technology. And I also was an avid reader of science fiction and Isaac Asimov was a favorite author which had lots of books of bass stories about robots and technology in AI. So when I went into college in the early 90s, my goal was to get an AI PhD in artificial intelligence. Interesting. When I say that to some people these days, they're like, wait a minute, it existed way back then. Yes, it existed way back then, even before then. What was interesting is as I was going through college, I got a job offer and it was like, do I take on more debt and go after that PhD or do I go out in the real world? I chose to go out and start working and worked my way through various companies in different positions as a software developer initially, but then moved into project management, product development. And in the early 2000s, I had an opportunity to start my own software development consulting organization. And at the time, I was working on a project in diabetes management and it was just fascinating. I worked on projects in many industries, railroad cars, accounting, inventory, and they're all essential to life. You need these systems. But they weren't very fascinating and healthcare just fascinates me. To this day, 20 plus years later, it still fascinates me. It's the last industry to really embrace the idea of data and decision making in that data. That's where the AI comes in over the years I worked on various different iterations of AI systems, my expert systems, going back into the early 90s, generative AI, not generative genetic algorithm types of systems, which is fascinating. It was one of my senior thesis at at a college. I don't use genetic algorithms to help answer some of these questions. So it has been around for a while and used in lots of different iterations over the years. Generative AI, which is what everybody talks about today, just happens to be the newest iteration of that technology, which does represent an inflection point. Yeah, absolutely. And I think I'm someone who's personally really passionate about AI. I think it's an incredible tool. And I think it's important to understand some of the risks. And I've been interested to find how much of a hesitation there's really been on the use of AI and healthcare. But I've only found it in the applications that I've examined with a critical eye, that it really eases the workflow and reduces unwanted workflow, especially for frontline caregivers, or enables high-level managers to be able to, instead of spending all their day auditing things, they can have the answers they need in minutes and then go solve problems. So why do you think AI is so important, healthcare? You mentioned the diabetes thing had really interesting outcomes. Maybe you can speak of an example like that, where it has just really made a huge difference, particularly on the frontline, but also in management. One of the biggest use cases coming out of the use of generative AI today is the idea of ambient messaging, note-taking. So listeners may have seen this just in using teams or any of the various software platforms out there around meetings where it takes notes for you. And that's a huge inefficiency that I saw over the last two decades, that the advent of medical record systems and just the overwhelming need to document everything that happens. Why did these medical professionals, healthcare professionals, healthcare get into this, is to work with people, not to take notes, not to justify their billing data. And that's where very quickly we're seeing an impact. So the technology, the generative AI can listen to the conversation in the room and then record notes and then later on they can be reviewed. What's fascinating by that is the idea that generative AI can hallucinate or make up stuff. And as I've looked at this and examined it, I've found all sorts of statistics of 9% of transcripts have a mistake in them. But what was fascinating about that is that's human mistakes. I was going to say I feel like humans make it at least that many errors too. I used to be a compliance officer, that's just fact. And that's the interesting part about this that everyone is like, oh, can I use AI to do this? And can you use people to do it? Yes. People are not perfect. Humans are not perfect. We've developed the processes to account for that imperfection. So while the industry is working out how to improve AI to be better than humans, let's apply the processes we have in right now to overcome some of those hurdles. So yeah, I think there's a lot of promise there in a number of different areas of just how do we make that difference. We were talking about a diabetes real quick too of how do you use AI? And this is where you get into more some of the machine learning aspects that the side of AI that's been around for much longer. And they're using it to analyze data in subtype to diabetes. So instead of just having type two diabetes, what are some of the underlying reasons for that? And analyzing vast volumes of data, the machine learning algorithms can output that information and find those patterns that you and I can't find because we just can't consume that amount of information. We use it in diabetic retinopathy. In diabetic retinopathy is a leading cause of preventable blindness related to diabetes. If you find it, then we can make a difference in that person's life and save their vision. Today, most of those images when they're taken and looked at by a human, well, computers and technology, machine learning algorithms can do that much faster and better. It's being used in breast cancer detection, where it's much better than an actual human looking at those images. It reduces false positives and false negatives by applying those machine learning algorithms. Yeah, it's really fascinating. I heard of an example recently where it was a note taking service listening to physicians in the office and a patient was talking about how they had pain and they were giving themselves whiskey every single night to treat the pain. And the EMR system interpreted that and documented it as resident self-administering with alcohol, including, and just like a physician would do it. And I was fascinated by that ability to not just copy down what's happening, but really extrapolated into the appropriate medical terminology and documentation. And we're recently connected with a home healthy EMR that does the same thing. It'll read the chart to you while you're driving to the patient appointments. So that's already saving 10 or 15 minutes per patient. It'll listen to the visit. It'll document everything to read it back to you. You can correct it with your voice. Again, taking advantage of the time you're driving in between patient appointments. Instead of that being lost time, and what really happens in home health is that a lot of these clinicians are working a full day and then they're going home and they have two or three hours of documentation. And we know the quality of the documentation when you might be sitting with a glass of wine and watching some television. So I'm really just fascinated by all of the opportunities and applications. So what do you think? I'm a healthcare executive. I'm a senior living home health or another setting in healthcare. What do you think are some of the key considerations we should have with AI and memory learning in healthcare? >> Yeah. >> Today you do have to worry about a couple key things, one, the idea of hallucinations. So you have to remain vigilant when you're using it. So you can't just take it and say, okay, here's the note and move on with life. You should do the review of that to make sure that it is accurate. So that's the first thing. Bias is the other thing. And I also find that. fascinating because we as humans are inherently biased. But if you were to ask your physician, care provider, if they're biased, they're gonna say no. When in reality, that's just the way the brain works. That's the way we categorize information. We do have inherent biases every time. And using the gendered of AI systems and how they're trained, there are biases there too, but you can ask it about its biases and how to better address them through that. And how do you make a difference using the gendered of AI systems to do that? As we're training the data in the systems, we have to be aware of that. And what's interesting, if you think about two different environments, so one, you have a home care setting, inner city, I live in the Philadelphia, greater Philadelphia area. So that's one environment, or you go to rural canvas, we're farm country, totally different experiences. In both cases, the people involved are very well intentioned, wanna provide the best care they can. But because of those differences, the AI that has been developed to work with the urban Philadelphia based environment may not work as well or as efficiency in that canvas environment, just based on how you're recording data and the information got it. So what's interesting is more and more vendors bring these products to market and they're being implemented. And the FDA does in some of these cases review these systems to make sure that they're okay. You do really need to do that testing and double checking of how does it work in your environment to make sure that it works for you. - Yeah, I think ultimately the biggest safeguard is just the human. I think we're nowhere near, nor will we be probably in my lifetime the point where we can trust AI to do this stuff without a human, still reviewing it, approving it, it's still my license on the line, it's still my signature on that paperwork, it's still the facilities who's responsible for everything that gets put into that documentation. So I think that really preserves the job workforce, but like we talked about really removes some of those unnecessary paperwork style type of things that we've got to do in our normal workflow. And a lot of people, even when you explain that to them, they still have a hesitation with AI. What do you think is a legitimate, the biggest risk associated with AI and healthcare and how can we mitigate that risk or some of the other risks that you hear about? - The biggest risk is, I think related to the human. And the fact that humans are fowlable, we make mistakes. There are hundreds of thousands of documented medical mistakes in hospitals. - Yeah. - Many other environments. And I'm not saying the AI is perfect by any means, but there's a reason for a second opinion. So how do we use the AI to augment the human? So not artificial intelligence, augmented intelligence. How do we use that to improve it? So I think the biggest risk factor in all of this is really the humans. And he referenced that you don't think it's gonna be trustworthy in your lifetime. I beg to differ. The pace that this is advancing, it's not that far away, where it is going to be better than the average human. I don't think it replaces humans. It doesn't replace the doctors, the caregivers, it's used in conjunction with them. And that's why I really like the idea of augmented intelligence. How do we make our lives better? That care management experience better, from an efficiency perspective, or just helping that patient improve faster. Think of all of the medical research that is published every year. I saw a stab that's like a million articles are published. You and I can't keep up with that pace of information. The computers, the technology can. So bring together that augmented listening, the idea that the AI can understand what's going on with you as a patient and as individual and has access to all of that cutting edge information. Bringing those things together and servicing novel, interesting questions to the care provider. Hey, why don't you ask about this? And I still today, as a society, we're very much like we need to have a human in that loop. So I think that those questions are surfaced to the care provider, which then can make an informed decision of like, OK, is that appropriate to ask this question? We're not there yet. But I think very quickly, we are going to be there. I think there's an I don't even think I know. There's an obesity crisis in America. There's a mental health crisis, if not in the world. And there are not enough professionals to provide the care and treatment to all of those people need. The genitive AI with the right guards and guardrails in place can make a difference in those areas. And it's only going to get better. Yeah, I think so too. In the late '90s, early 2000s, when I would say to my friends, I loved watching Rosie the robot on the Jetsons. And she cleaned the house for them. And I said, I think that's going to happen soon. And they're like, you're crazy. And now how many of us have robot vacuums and robot mobs and things like that? And there's more and more of that even being created in healthcare. There's a robot right now. They're working on that can provide bathing and dressing for residents with dementia and things like that. So it's exciting to see what's coming. But as a senior living or home health executive, if I'm looking into getting AI into my organization, what are some things that I can do to provide reliable and accurate-- make sure that the AI is providing reliable and accurate recommendations. And is really flowing well within my team. I think the same thing that you would do for any other system that you're going to implement as part of your processes, whether it's administrative or care management, you want to do those same things. So going into it, having clear expectations of what do you want the outcomes to be, that you do have the processes in place to monitor the technology to make sure that it is responding appropriately. And when it doesn't, that you have the processes in place to mitigate that, to identify it and, in course correct, as necessary. So yes, it is a novel technology. It is pushing the boundaries of what anybody thought was possible. But it really comes back down to the same good sense that we've all applied to various different health technologies over time. Yeah. So I've heard that AIU even referenced this earlier that it can serve as a second opinion for doctors or other health care professionals. What does through how that could work? Yeah, and I think that's a great use case. My mother-in-law usually go past away from Lupus. It's still to the way of condition that is hard to diagnose. So as that second opinion, if a physician that might not had seen those cases, if they may have seen it in medical school, but if they're now out in medical school for 20 years, 30 years, and haven't seen this, it's not going to be top of mind. So by talking to the generative AI systems, the chat GPTs, the clause, whichever one you use, in a hipocompline fashion, and we'll come back to that real quickly, but in telling it about the situation and simply asking it, what do you think this could be? And what's fascinating about these AI systems is if you're not familiar with using them, if you're not comfortable using it, you can literally ask it. Ask me-- good, Chachapu. Ask me questions so that you can form a better opinion. And ChachapuT will ask you questions to help figure out how to respond to that. So there's this whole idea of prompting that's come around recently, how to get the best out of these systems. And that has validity. It's good to know and understand that. But if you simply ask it to help you do better, it does, which is, I think, a fascinating way for using it. There is a lot of reference hip-a, and there's a lot of concern too about hip-a, and these technologies. But you don't have to go into ChachapuT and say, hey, I'm seeing patient Bob Smith today, who's 51 years old, and lives at so and so address, and is so security- and he doesn't need any of that information. You can generically talk about the situation and what you're seeing as an individual to be able to get that second opinion. And also, the eventers that are using these systems, like the Chachapu OpenAI, which is a company behind ChachapuT, they will sign business associate agreements with you today to help from that perspective too. So I would also do that just from a cover your basis kind of situation to make sure you have the appropriate agreements in place. Yeah, that's really fascinating. I didn't know that OpenAI would sign a BAA. So that's a great tip. I'm going to take away from this conversation today. I use it all the time. I'll be doing audits for a client. And there's one client in particular. I'm helping them with some of their nursing documentation errors. And I was getting to the end. of the report and saying how many errors that they have in what dates over the last month that they have the most errors. And I thought if I did this by hand it's going to take me hours. I put it in chat GPC with the patient information removed and I had my answer in a matter of seconds and now we have much more actionable data to be able to say wow it's you know 75% of these problems are happening only three days of the month. Now we can go and root cause analysis. Why is it happening on these days? Are there particular people that it's really a problem with and you know dive down deeper and actually find solutions. And if something I probably never would have taken the time to do on my own because it just took too much time. So yeah I did go ahead. I was saying and you can have it graph that for you and chart it. So when you have to do that presentation to the management to other you know organizations it just made your life that much faster and easier to be like here's the pie chart here's the graph to be able to present that in a much more meaningful fashion. We always talk about internally talk about telling a story. Well you know that picture definitely helps tell story. Yeah one of my co-workers uses it way more than I do and he's been teaching me a lot about how to get the most out of it and one of the things he'll do is he'll kind of say to it I don't like the personality you used and that in that answer I need you to be more like this. And the answer really adjusts accordingly. I mean I'm amazed that you can talk to it like a person with a personality and that it will actually make those adjustments. So I really encourage everyone on our leadership team and I encourage any leaders out there to really embrace it. Give it a try see what it can do for you. Test it's boundaries because you're right there are hallucinations but I've yet to have one of them happen in any of the work cases that I've tried it for. You know I've just consistently found really clear and actual information and then it has the link you can go check the sources and it's like okay it's exactly what's going on but it's just very succinct and very helpful. So if you're someone who's a skeptic I'd encourage you to just give it a try a few times and see what it can do for you. That's absolutely great advice and do the same thing because as you know I'm out there in the world and talking to people about it and they're like yeah I'm not too impressed. Like you didn't use it long enough and simply ask it to respond like Bill Gates or Bill Murray and you're going to get completely different responses and there's those sort of aha moments I did I was playing around with it a while ago and it was like creating an application to help people get over the fear of peanut butter being stuck to the roof of their mouth. Something just completely off the wall kind of thing and its response was are you sure? That's unusual. Totally called me off guard. I didn't expect that at all but it noticed that okay this is outside of the usual you do really want to do that and I said yes and then it provided me with a lot of useful information but it's just fascinating to see that. It was very you know yes it's not conscious it's not a person but definitely amulets it. Yeah it's definitely happening. So you know what do you think the future of digital health is going to look like? What do you think is going to happen in our lifetimes especially if we look at things like wearables remote monitoring. I mean I hesitate with wearables because a lot of patients aren't compliant especially when you most need them to be so you know especially on the remote monitoring I'd love to hear your thoughts and feedback on that. Yeah there's a huge potential for opportunity here still and it's getting better and better and better and what I think is fascinating by the wearables or just more generic monitoring technology is it's fading into the background where you know to monitor your heart rate you have to wear a strap or you know around your heart it was really difficult. Now you can wear your ring your apple watch but there is technology that just looks looking at you through like the lens of the camera can figure out what your heart read it can look it and figure out your blood pressure and all sorts of other metrics. So you as an individual you know you get up in the morning and you go to the bathroom and you look at the mirrors you're brushing your teeth and you can get health information from that individual. So as you think about you know different living situations as our population gets older and older kind of thing where do we want them we want them to age and place it home with their comfortable and do they need resources and and that technology is is there today and it's only getting better. So as you think about how the AI is useful it can understand the data pattern so my okay my mother got out of bed this morning she used the you know facilities she turned on the coffee machine she opened the refrigerator she turned the TV on her normal pattern of behavior well okay tomorrow it's not at the time mom didn't get up she didn't start her coffee that's unusual let me place a call and you know see what's going on kind of thing maybe she's not feeling well that day well okay the technology can even like externally you know understand you know the different rings it's like my temperature is elevated today are you feeling well yeah incorporate that information and now I don't have to worry about my parents at home and they're much happier because they're at home so I think there's good I think there's tremendous potential in in the idea of wearables and and sort of that external monitoring that you don't even have to wear something that is tracking all this information and the generative AI what I find interesting about it too is makes the information that you impart to the individual or the caregiver much easier to interpret and can meet that individual where they are in their health journey and their health literacy their numeracy if they don't understand you can and can you explain that more simply to me I have a sister that's studying quantum physics way beyond anything that I can figure out but it's like explain that to me like I'm in kindergarten and it'll come up with like an analogy that works for me and if I don't understand it I'll be like come up with another one so it's fascinating potential yeah I think one of things I like the most about more of this passive remote patient monitoring whether it's activity or vital signs or things like that that require no real compliance from the patient other than the initial setup agreeing to initially set it up is I think it really gives people a lot of privacy I think that's the thing that a lot of elders really have the biggest trouble with and I mean if I'm in law to you I wouldn't like to see there that a nurse has got to come to my house every single day or a caregiver's got to be there every single day and really it's not because you can't take care of myself it's just people are worried I might fall and no one would know about it and so if you put these systems in place nobody has to come see me unless I want them to you know and I have the opportunity to open my home to the people I want to open it to rather than having to just let people in my home all the time in order to have appropriate monitoring or have to be able to get up and leave my home to go to an office to get a blood pressure check and things like that that used to happen so I really like the idea like you said of aging in place but also how much privacy that it actually gives to patients finally what is something that I didn't think to ask you because you know a whole lot more about this than I do that you think our listeners need here we did talk about it but I think it's worth empathizing is the idea how to get started with AI and it's really don't be you don't need to be intimidated by it because it's just you in the machine sitting there talking chatting kind of thing and simply ask it how to do better teach me how to use you the computer better and it will do just that so it's yeah if you're not impressed then you're not you're not trying hard and honestly it does I don't think it takes that long or that hard to be impressed by what it can do today let alone what it's going to do tomorrow you know it bill gates said a while ago that we as humans tend to overestimate what's possible in two years but underestimate what's possible in 10 years I don't think we understand where this technology is going in 10 years I don't think we can conceive that let alone two years is probably interesting too because you know I think of this is up there with you know the Gutenberg press you know like book printing you know electricity the internet you know smartphones these these technologies have made significant impacts on society they're coming faster and faster which makes it challenging today but um just hope everybody uses it for good that's right that's right well I can't thank you enough for joining our podcast today really appreciate you sharing your expertise and your insights as always if you'd like to learn more about RJ, Asthena or gravity you can always find us on LinkedIn and don't hesitate to reach out to me directly thank you for joining us and if you enjoyed today's content don't forget to subscribe to our podcast remember it's not just what you know but how you apply it that makes all the difference see you next time