Listen to Dipti Gogate on AI in healthcare

Transcript

How are you doing today? Very well. How are you? I'm doing well. The weather's turned here. It was really hot and sunny and you know, we were looking for some respite and we finally have it in rains, so all good. Couldn't complete. And I think there must be some sun out at your end as well. Yes, there's about to be a little cloudy this morning, alright. Wait, you know, before we go ahead. Thank you for you know sharing your time. Do appreciate it. Before we go ahead, I would quickly like to introduce myself. My name is Dipti and I work for GRG Health. Currently I have about over a decade of experience in market research. What GRG essentially does is you know gathers actionable insights for our clients so that they can make business wise decisions. So their ROI's are great basically. Spelling out data for them that they would otherwise have no access to. Right. And you know, for the benefit of our listeners, I would also quickly like to, you know, give a short introduction about you. So could you just help me with it? You know, could you just tell what have you been up to for the last few years? I've certainly. So I have been working in healthcare now for about 30 years. First in. In management roles, healthcare administration, management roles, not an IT and then and then an IT have been an IT for about 20 years now and then the last six years as Chief Information Officer at Dayton Children's Hospital, which is a small health system in the Midwest in the US. And so we are on Epic for our EHR very relevant to this conversation. We're on work day for our ERP. We're on, we're Microsoft Shop 4200 employees. So, so we're not a small organization. Terms in the US where a small health system, right. I also see that you've got some military background in as well. How many years did you serve? Yes, I'm I'm a retired from the US military. I was seven years in the Navy doing not healthcare doing on ships in the Navy and then 13 years as a healthcare administrator in the year. Great, great to know about you and thank you so much. Let's go ahead. And begin, you know our quick questionnaire. So my first question to you today is going to be how is, you know, according to you, how is AI reshaping the landscape of healthcare and what are some of the most exciting advancements that we've seen so far? You happen to mention Epic earlier. Would you say that you know Epic is by far the most popularly used or are there some other softwares out there as well? Sure, that's an important question. So let's start let's start with that Epic is in some ways far and. And the leader in some ways not it sort of depends on the the type of hostiles of health. So in if you measure it in terms of hospital, Epic is approximately 50% of all the hospital bed in the US. However, if you look at it in terms of the health systems, it's about 30. In other words, it's in tends larger health. It is not affordable or practical in the smallest deposit. So in rural areas, U.S. hospitals tend on Meditech, Pop EHR and then you also have Oracle, Cerner and sort of in the middle there. So, so, so, so Epic, it's fair to say that Epic dominate large health system down to about the size of my. So would you say? That, you know, in rural areas Meditech rules because it's cost effective or you know it's epic has difficult to reach accessibility. What would be the prime reason for cost effectiveness is the is the main thing many of these hospitals are not only on MEDITECH, they're on very old versions of MEDITECH. They're they're literally on the verge of bankruptcy all the time and so they can't spend any money on really any IT wise they're just. They're just trying to survive. Alright. And you know, going back to my previous question, why, you know, what do you think is the most exciting advancement that we've seen so far? Sure. So first of all, we have to start out when we talk about AI, we have to be very clear what kind of AI are we talking about? Are we talking about generative AI, obviously the new stuff or are we talking about predictive algorithms, which pending on your definition of AI? Every I that's being used, so it could be. Epic, it could be EHR's right. But the reason that distinction is important is because the predictive algorithms are not new. They've been around for a very long time, right? And, and when we started using machine learning to build better predictive algorithms, you know, there, there's all sorts of applications of that. We're not done with that. It's not like we got done with that. And then and then generative AI was invented, right? There's still more of that to do. However, most of the time when people are talking about AI now, they, they really, they really mean generative AI. The reason that we have to separate those things, it is because they are completely, completely different and handled differently for their predictive. Algorithms, let's just leave it at that and call it a predictive algorithm, OK, going to predict something, I'd have a bunch of data. I did some machine learning, I built a better algorithm than we had before. And now we want to use it somewhere in the in healthcare, maybe on the clinical side, maybe on the business side, right, right. Well that we can measure the accurate accuracy of that predictive algorithm and then we can say it's, it predicts something this much better than the last one and it's going to cost this much to implement. And therefore, it's smart to do or not not smart, right? Oftentimes you can have something that predicts something slightly better than the last thing, but if it's gonna cost a lot of money to implement it, it's just not worth it to OK, so there's a lot of that out there for for generative AI. The challenge, of course, is with hallucinate generative AI, we have to be very, very, very, very. And you can't just start general AI to dying patient or so. So there are powerful solutions that are coming. But the ones that are being put into place now tend to be the more back office or efficiency kind of things. The big one that everybody's talking about, most health US are seriously looking at if they have not already purchased is the ambient AI digital Scratch. So companies like so Microsoft has a solution here. The DAX Nuance, DAX Copilot, a BRID is another one. Those two are the ones that have the best epic integration. The most epic customers are. Sit down. And there's also, there's other good ones out there, Nabila. Suki And so this is where the doctor can doctor walks into the exam room, puts down their phone, says, hey, since the patient AI is gonna help me document my note for today. And this means that I can spend more time looking at you and talking to you instead of you seeing the back of my head while I'm typing in my computer. And this is going for the early adopters. We are about to adopt. You have not yet, but I'm paying a lot to adopt and it's pretty overwhelming how popular with doctor. Really saves them time. It saves them enough time every day that it really makes a difference to their, to their practice and to their, to their well-being so they, we can hopefully avoid, you know, physician burnout. And I believe it also opens up some space for them to actually have a meaningful conversation with their patients to get to know them. Yeah, better and maybe, you know, understand the pain points that they're facing. That's great. And with any new technology, there's challenges to adoption. You do have to you have to speak. Am I a little bit you have to say it now I'm now I'm looking in the left ear. Now I'm looking on the right ear because AI doesn't know say that right. You have to introduce the people in the room, the Pediatrics. That's important because of course there's oftentimes the patient is not the person that the doctor is actually talking to talking to mom and dads. Yeah introduces in the room or of AI. Yeah, true. So you know, if we talk about trust, like you already said that you know it it's really popular with the doctors because it opens up the way they treat their patients and you know. Gives them more time because there's nothing running at the back of the head. If we talk about trust, how do you think, you know, patients and healthcare professionals alike feel about and trusting AI with their health decisions like, you know, are we ready to hand over the reins or is there some kind of skepticism still in the background? Well, there's there's a lot of skepticism about. So what I said a minute ago, we're not letting the generative AI just diagnose people right for lots of for lots because nation. Because of malpractice gonna get sued when the AI is wrong right the the American Medical Association came out not long ago and said well the AI is wrong. It's not the doctor's fault it's the AI's fault. So the developer of the AI but that we're not ready for that either. So I still have to hold doctors accountable for their for their practice right. That's that's the way the system works. So there's lots of that is not have real soon now what what clinical leaders that are very smart about general AI what they like would like to say. Let's not use the term artificial intelligence. Let's use the augmented Intel. They'll call it AI clear it's helping doc is not replacing doc. And so for example, you could have some, some AI developers are working on tools related differential diagnosis. So that would help a doctor with a diagnosis and perhaps suggest, hey, I saw this lab test or something in the thousands of data points in in the EHR that might suggest this. Rare disease that you hadn't thought of in your differential diagnosis. Maybe you should add a order another lab test or ask some more questions. Ask the patient some more questions. But to rule out that where maybe the AI could augment Vitor with that information, but still that's not actually die, it's supporting or suggesting things that they might not have passed through the doctors based on the data. The way I can do very well is just summarize lots of data, right? So take the case of of specialists that are a new patient, so and on college for for cancer care, right? That typically has lots and lots of incoming information when they're going to see a new patient. And it would be really nice summarize that AI to help them be more efficient because they if if AI can thoughtfully summarize thing that the doctor, if they had enough time, could read all read all of that. But I can help summarizing. That's, that's true. And they it could give them, you know, a distinctly more clarity than ever before. So, you know, while we've established that, you know, there is no way I can possibly replace the doctors. And you know, it is going to be always used for more of, you know, summarizing data for them, which will in turn cut down their, you know, looking at each patient time. Would you say there are any potential pitfalls or ethical dilemmas that we need to watch out as AI becomes more and more? Integrated into healthcare sure. So some of the ethical dilemmas include their their yeah can be and there has been proven to bias in you know in some of the now I'm now I'm going back just predictive algorithms. But this could also generate is you know if you have for example more well known examples if you are using the healthcare data that's available and the healthcare data is available use towards people. And I'm now talking in the United States because we do our health insurance here very different resolution not smart we do it. Is that it's the people with insurance? Well, that so so you have you have a less data from the people that don't have insurance, which tend to underserved people tends and there's some racial, you know, bias there. And so sometimes you built the algorithm with not a representative of a patient of their some biases in the data. That way. You have to be very careful about that. Another thing that we have to think out think about is access to care. So in terms of an ethical dilemma, so let's say. That the algorithm is slightly biased in some way. We, we maybe even know it's slightly biased. But that capability could be, or another example with the generative AI that maybe there's a maybe 1% of the time it hallucinates, right? We, we, it's not perfect, right? But what if that because it's AI and there's not enough doctors and maybe we're talking about delivering care in underserved area. What if that's the best care available to that patient at that time because they weren't otherwise? Going to be a doctor then what it then there's an ethical concern about if you withhold that it's unethical hold that yeah, right. So there's all sorts of effort there's all sorts of considerations. Alright and so how would you say are the healthcare organizations navigating the challenges of, you know, integrating AI technologies into existing systems and workflows because, you know it's always difficult to introduce a new system into the place and what kind of strategies are they using and what's proving to be most. Effective sure. So now this is where Epic comes in. This is very important so Epic has had a long time partner Michael over in a very very good of all the eye stuff just Epic is building into Epic all sorts of new generative AI features that have Epic customer can use if they want now it's important to know that Epic is really a big development platform. So when you take your upgrade from Epic, they they do quarterly. Upgrade some Epic customers upgrade every quarter or some like us upgrade twice a year, take 2 upgrades twice a year. But when we get new functionality available to us and maybe we have 100 new feature, maybe we only turn on 50 of them because I don't know where children's House of them are not useful treating children or whatever it is. And so some of those new features now are powered by generative AI. Yeah, that's a good, that's a good news. There's cool new things in there that are powered by general AI, but they may or may not be worth our time. Are worth our money to implement it just depends what they are So what what a lot of people are doing is letting other letting other people go 1st and see how that new AI feature that epic offers this is that truly delivering return on investment you know because there are there is also a growing trend as people are starting to publish on all of this that you know it's it's hard to get it it's hard to get ROI on some of the.