00:02.09 Ratnadeep Bhattacharjee Hi everyone, and welcome to another episode of Leaders Perspective, where we bring you in insights from forward thinking minds and founders shaping the future of healthcare and technology. 00:13.56 Ratnadeep Bhattacharjee Today's episode is titled From Data to Clinical AI, and we are diving deep into how healthcare organizations can drive real, measurable innovations using data governance, AI, and Gen AI in clinical settings. 00:29.28 Ratnadeep Bhattacharjee I am joined by someone today who has been at the forefront of these transformations, right Chris Hutchins. Chris is a highly respected healthcare data leader with expertise, in fact, deep expertise in enterprise data strategy, governance, AI, ML, and scalable self-service analytics. 00:48.39 Ratnadeep Bhattacharjee He's also are particularly excited about the real world potential of Gen AI and LLMs in improving clinical workflows. Chris, it's a real pleasure to have you here. 00:59.43 Ratnadeep Bhattacharjee Thanks for joining us. 01:01.83 Chris Hutchins Well, it's a real pleasure, and I really really do appreciate the the invitation. excited to have a great conversation with you this morning. 01:10.92 Ratnadeep Bhattacharjee Great. Chris, to start us off, could you give our listeners a quick overview of your journey? You know, kind of how you found yourself working at the intersection of healthcare, data and AI? 01:25.97 Chris Hutchins I would love to. Thank you. Thank you for that. I don't want to go back and and give you the you know the cradle to this point in my life, but i have been around health care since I was really young. 01:38.96 Chris Hutchins ah I grew up in ah in a home there was associated with the area we lived in. my my mother worked in a hospital. My father worked at the the at the the college. and this was around Dartmouth College up in New Hampshire. 01:51.73 Chris Hutchins um And I just got to be around the healthcare care profession. When I was old enough to do part-time work, I got to work in the hospital a little bit. That was my first exposure that got me to realize that this is this is a mission that I really felt connected to. 02:07.22 Chris Hutchins ah I was working in an x-ray department and right across the hall was the ER. r So I saw enough to know that, you know, I don't think I could be a physician. 02:18.45 Chris Hutchins I think I don't have this timing for this, but the, the amazing opportunities that I had, the wonderful clinical leaders, physicians that I got to see, even at a young age, I really fell in love with the profession because I saw their faces light up when they come face to face with an individual that they're, they're trying to help. 02:37.34 Chris Hutchins And I've never been able to shake that connection in the pole to find ways I can plug into that. um There's ah certainly a length a lengthy evolution. i think my the early days... 02:51.62 Chris Hutchins There were some technologies that I thought were really, really cool and exciting. We were actually using magnetic tapes to do claims and ah transactions back then, these big, massive things. 03:03.48 Chris Hutchins ah you know they The transition to PC-based applications was something that occurred. I don't know if it was the... The organization was the first that did it, but the guy who was actually installing the software, he came out in my office for a week. 03:19.13 Chris Hutchins So we we know influenceation implementations don't take that long and these days. ah But... Over time, I just had, you know, really great opportunities to learn and grow. 03:32.23 Chris Hutchins um and And for anyone who's out there trying to encourage their their their young ones ah had it how to really go about things, curiosity is amazing. 03:43.85 Chris Hutchins um Just trying to figure stuff out and figure out how to apply it to make things better and more efficient was something that came kind of natural. But I saw my dad learning using FileMaker Pro. He transformed the way the services were run and where he was working at in the college. 04:01.66 Chris Hutchins That kind of inspired me. Yeah, data's really cool. The database stuff. um want I want to learn this. Um, but all of that opened up doors and the sequences were just unbelievable for me where I had opportunities to learn, be exposed to more technologies, more people that would, uh, freely give advice and help me along the way so I can understand how to do the things that that are necessary. And it just created all kinds of open doors one after the other. 04:28.22 Chris Hutchins Um, for a number of years, I was at one health system up there in New Hampshire for 15 years. Then I had an opportunity to go to to work at Mass General but in Boston with some extraordinary leaders there, um sitting and in rooms with people whose names are known around the world for what they do, you know, in the healthcare care space. Unbelievably brilliant of clinicians and leaders. 04:54.50 Chris Hutchins um And learned a ton there. And my boss, who is the chief financial officer, really was passionate about data and analytics. And he got me really, really more excited about that than, than I think anyone else has. 05:10.24 Chris Hutchins um but there was a vision that he had and that again, opened a lot of doors and over the course of the last 20 plus years, I had the opportunity to be at that Mass General Brigham port was, was Partners Healthcare care at that time for about 15 years. And then, 05:29.91 Chris Hutchins We had just finished our Epic implementation up there. We consolidated three really mature data warehouse environments into ah an enterprise platform. And then suddenly there was a phone call about an opportunity to go to New York. um And my my first thought was, this is probably about the most concentrated and diverse population center anywhere in the world. 05:57.07 Chris Hutchins if the data can be harnessed, that's sitting there, there is no more potential to have global impact than you can have in that kind of a marketplace. And so for me, that was a really big deal to be able to go there and be part of a team and build a team around me that could start to do some of those things. And i don't need to do run, run through the gamut, but there's the things that led to the success of the organization through the pandemic that, 06:25.27 Chris Hutchins was it a testament to the to the great leadership in the organization and they they really decided to to invest. and I was able to have some of the most brilliant minds around me to help me all the way through it. 06:39.03 Chris Hutchins And my only regret from that my days in New York is that I couldn't steal the guys that I got to work with every day. they They're just great, great people, brilliant and very passionate. 06:52.02 Chris Hutchins um But the the last you couple of years, i had the opportunity to come to Nashville and but been working in a national health system that is in the for-profit side of it. 07:03.89 Chris Hutchins which is quite different from how a nonprofit operates. And I learned this. um but Again, great experience with with some exceptional leaders. And I still to this day cannot figure out how one person can possibly have this much good fortune and and opportunity um to say that I'm blessed is probably about as understated as I could communicate to you. 07:30.48 Ratnadeep Bhattacharjee Wow, Chris, two things, right? Two things that, at you know, I can say. First thing is, what a journey. Second thing is, it seems like you are full of gratitude for the journey that you have had till now, which is which is honestly very rare. 07:45.63 Chris Hutchins Yes. 07:45.90 Ratnadeep Bhattacharjee But, you know, lot I guess our listeners will love what you have to share today. So looking forward to it, Chris, you know. One thing I've realized when you were kind of giving... 07:58.36 Ratnadeep Bhattacharjee Your introduction, right? You have you've kind of worked with some major, major health systems and, you know, health care organizations. What, in your view, defines a mature data strategy in today's health care environment? 08:20.95 Chris Hutchins I think that the thing that's so profound to me is that early on, we've started we just basically started catching up with what our eyes used to be able to tell us, which this seems kind of funny, but I was at ah at a data event one time and I had a chief data officer sitting next to me that was talking about the fact that having a lot of information volume in your in the electronic health record was a predictor of sick, you know, readmission risk, things like that. kind of chuckled because when I was working in the x-ray department, when the orderlies would come to bring a patient for films, they would put these stacks of paper records on the counter. 09:09.18 Chris Hutchins I'm like, oh my gosh, without our technology has caught up with the what our eyes could tell us before. Like, this is crazy. But the the the way that things have evolved is really remarkable in many ways. But in some, it's almost that we do too often still what I think we did early on. 09:34.62 Chris Hutchins We were trying to solve for something because it it was exciting. The technology was there. The intentions were there. no There's no question about that. But back in the day, if you went to see your doctor, 09:47.25 Chris Hutchins while While you're in the exam room, they probably were using a dictaphone to record notes in their observations. um Either after that particular encounter or at the end of the day, they would hand their tapes to a transcriptionist who would go, type all this stuff up, putting it into a report. 10:05.09 Chris Hutchins It would come back to the to the physician to be reviewed and approved, and it would go on the into the paper record. um Our intention was to make things better, but we replaced the wrong thing. 10:17.44 Chris Hutchins We replaced the transcriptionist and made the doctor start doing all this documentation, but he's doing it and on the computer. and That was one of the things that to me, it was like, that was kind of the first step in a whole evolution of things that have caused whole bunch of layers to be added. 10:38.76 Chris Hutchins onto the folks that are responsible for clinical care, whether it's physicians, nurses, or even a lot of the clinical support staff, there's things that they have to do now because we've improved their lives with technology. 10:52.18 Chris Hutchins um But now we're at a point where we're starting to realize the technology that we have available to us can actually reduce those burdens. And i think that's probably one of the most profound opportunities that we find ourselves in right now because there's never been a higher demand for clinical care than there is right now in the behavioral health space in particular. If you think about what's happened over the last eight years, we're already on a fast track to crisis, that honestly, because of the influence of social media and a lot of other types of value-based issues that we experienced as a society. um The pandemic just put that, yeah li it literally put, li 11:37.53 Chris Hutchins basically strapped a rocket to it and it was driving hard into a wall. um There's never been a time that we've needed to figure out how do we free up our physicians and our clinical teams to really do the things that they went to school to do and not burden them with all this other stuff. 11:57.34 Chris Hutchins That's where people like us, I'm not clinical, but I can figure out, you know along with you, We will find those areas where we can eliminate things. We will create time for them to be face-to-face and present, you know, and remove friction from those relationships. 12:14.34 Chris Hutchins But I think that's probably the biggest need that we have at the current time. And the challenge with it is there's so much concern about the ethical use and responsible use of these technologies, and rightly so. 12:26.72 Chris Hutchins But somebody has to be responsible for finding the way to say yes. What can we say yes to? right Our legal and compliance professionals, our IT folks, our information security team leaders, they're extraordinary. 12:44.47 Chris Hutchins And so I'm like, all right, I don't need to worry about that. um I'm going to make sure I feed them whatever they need. But I've got to find those ways where we can engage with our frontline people who are running operations, whether it's logistics or or if it's clinical. 13:00.68 Chris Hutchins I need to help find those areas where we can say yes, because there's tons of those opportunities that don't have the risks that everyone's so worried about. If you want to talk about quality reporting, it's mandatory, right? 13:14.09 Chris Hutchins That's a perfect scenario. I'm not trying to diagnose anything. I'm looking in the notes to figure out is there presence of or absence of a numerator and a denominator. And a clinician knows exactly what they did. 13:27.49 Chris Hutchins and if you put that information in front of them to respond to it, they'll know right away if we got it right or not. So these are the types of things where we can apply this and make a huge, huge difference. so And I think it was Dr. Eric Topol I heard about two years ago, he was just challenging us to figure out, can we take 30 seconds of burden away from the in any in every encounter for the physician? 13:52.04 Chris Hutchins Just that alone. it just removes some some unnecessary burden. But, you know, he in his own efforts, he was identifying one thing at a time. 14:03.96 Chris Hutchins And he found that as he kept doing it, he he he could actually see some pretty significant improvement in the capacity and allowing the physician to have a more seamless and frient frictionless encounter. 14:17.70 Chris Hutchins um it's It's just a really big opportunity and probably more needful than almost anything else. And you know if we if we can figure out how to start with where the pain points are, we'll have all the support we need from from these brilliant minds. They will be very happy. 14:33.64 Chris Hutchins They might actually get to go home and have dinner with the kids once in a while. but I think we we owe them at least a good hard run at trying to make that happen. 14:43.13 Ratnadeep Bhattacharjee Chris, I couldn't couldn't agree more, right? You know, this has been one of the most burning issues, you know, physician burden, clinician burden. you know These are the terms you keep on hearing and, you know, how agitate you know as much as you agitate the problem, it it's not enough. 15:00.16 Ratnadeep Bhattacharjee It's never enough, right? So I agree with you, right? 15:02.66 Chris Hutchins Yes. 15:02.71 Ratnadeep Bhattacharjee But one of the points that you touched upon, Chris, is very interesting to me is, you know, responsible AI, ethical use of AI, you know. So one of the key terms that comes to my mind is governance, right? 15:16.69 Ratnadeep Bhattacharjee When it comes to governance, how do you strike balance between, know, protecting data, which is of utmost importance, especially in a healthcare care setting, and at the same time doing some agile innovations, right? 15:17.78 Chris Hutchins yes 15:35.01 Chris Hutchins I don't know that there's a perfect answer, but I can i can share how I think about it. the the The challenges that seem to be the ones that I hear most are around risk for hallucination, for example, or you know, missing miss, missed, missing something completely, ah in, in the assessment of, you know, when you're trying to diagnose something and I agree, yeah, those, those are big risks. 16:03.94 Chris Hutchins But when you're talking about the idea that you already have errors, sadly enough, we have errors. 16:15.39 Chris Hutchins Um, audits prove that. And I don't want to, even go down that pathway, but but human beings are are going to make mistakes. And this happens to eat to the best, with the best intentions, the best training. 16:28.03 Chris Hutchins So we need to really be honest with ourselves as we're evaluating things and say, well, you know what? We already have inaccurate results from time to time because it's still human beings doing it. 16:42.24 Chris Hutchins AI can improve it. but it doesn't completely eliminate the risk. But I would argue that if we can significantly improve the odds of success using AI, then it's worthwhile to do that. 16:55.21 Chris Hutchins The balance thing that we need to do is to figure out how do we guard against really two big things. One is unintentional bias. And in the healthcare sector, i so I would suggest that that's what we're talking about for the most part. 17:10.55 Chris Hutchins It's just not that people aren't out there trying to cause harm. The Hippocratic Oath is very serious to to those who take it. And, you know, I feel like i owe a debt of commitment and dedication to find ways to allow them to operate within those parameters, right? 17:28.79 Chris Hutchins And so things like protecting against human nature, If you see something get automated, as a human being, after a while you start to trust it, you can trust it a little bit too much. 17:43.46 Chris Hutchins And so how do we make sure that we're providing the right boundaries and um tools to monitor? what things could be a problem. um You might actually spend three and a half minutes per note in a review process. we've useded We fast forward, we've got ambient listening going great. 18:04.14 Chris Hutchins We're getting great results. We're excited about it. the the The whole encounter is like summarized and I just take a quick look and yep, it's all there. It drops, it gets coded the bills go out the door and everything's great. 18:18.06 Chris Hutchins You take three and a half minutes and this Hutchins guy is taking like 45 seconds. 18:26.12 Chris Hutchins do you really have a deeper um look and approach to assessing it than I do? And is it the right approach? 18:36.85 Chris Hutchins Or am I just like super comfortable more so than I should be but with all the technology that you have so kindly put, put into my hands that I start to get lazy. And it's a, it's a human nature thing. It's not to say that anyone's thinking about doing this. 18:52.10 Chris Hutchins It's just something we have to be aware of. So there's, number one concern I have is what's missing, right? We don't know what we don't know. 19:03.87 Chris Hutchins And that's concerning if you're a physician facing a patient who needs to, they need some help, whatever their condition is, they that they really are hoping that they're going to get the best, the best care and the best result. 19:18.82 Chris Hutchins And for a variety of reasons, we do have a lot of risk around missing information. Um, I'll say this cautiously, but think about patient safety risk. This is one of the things that I always look for, any anything that might even resemble that a little bit so that I can help support the clinicians in advocating for what they need, if it's technology or whatever it is. 19:42.99 Chris Hutchins But if I went out and stood in front of a camera, in front of the media, and started talking about the need to have a patient a national patient identifier, 19:55.90 Chris Hutchins it would not go over very well. I would probably have a lot of people having a lot of nice not so nice things to say and questioning my my motivations. 20:06.55 Chris Hutchins and All this is the dynamic that we're living in, which makes this even more complicated. But what we don't know, great example, everything that we saw at the beginning of COVID, it looked like something else. 20:20.38 Chris Hutchins all these different symptoms, like it could, I mean, some sort of a pulmonary thing. We kind of figured that out eventually, but we learned a lot in a really short period of time, but the assumptions that we could have made could have been even more detrimental than they ended up being. 20:36.49 Chris Hutchins And so it's that risk. What I don't know, how important is it? And this is not a risk AI fixes. This is an existing risk that every encounter has, right? 20:47.82 Chris Hutchins around it. This is a real risk. And if I don't feel comfortable to tell you that, yeah, I know i'm not supposed to smoke, but I smoked two packs yesterday. You think you gave me great medical advice and you're like trying to figure out what else could be wrong. 21:03.43 Chris Hutchins shit It's just that I'm not, I'm and too embarrassed to tell you that I i haven't taken your instruction. <unk> There's just so many different components and ways this can go south. So the missing information to me is the big thing. 21:15.78 Chris Hutchins And then really making sure that we have the the right guard rules to protect our ah clinical people. They're going to use this capability. We have to have the right mechanisms in place to be able to monitor things, not in a big brother kind of way, but just to make sure that we're guarding against our our own and over trusting or I mean, sometimes you can get complacent, you know, if things seem to be easy and you're getting great results, but it's on people like us to figure this out. 21:43.89 Chris Hutchins We just can't keep adding that anything more to the physicians. Just make them comfortable. Yeah, we got this. We're on it. We're going partner with you. We're going to figure it out and make it better. But we're going to start by asking you where your pain points are, not coming to try to convince you about to use this really cool solution we just developed. 22:04.74 Ratnadeep Bhattacharjee Wow. I mean, ah Chris, some of the things that you mentioned, right? Hallucinations, bias, missing data. These all are, you know, phrases and words that are constantly being associated with something that you are particularly interested in, right? 22:25.18 Ratnadeep Bhattacharjee Chennai. LLMs. 22:27.04 Chris Hutchins Right. 22:27.07 Ratnadeep Bhattacharjee These are the buzzwords and these and the phrases that you mentioned are at the peak of you know discussions amongst developers, founders, you know people from the ethics department and whatnot. What ah what according to you, is getting you most hyped about Gen AI in clinical settings right now. 23:00.00 Chris Hutchins You just threw me the biggest softball of the day. I'm excited about this. I have had some time over over the last couple of months just to really play around and in find find out just how far I can push GPT capabilities for myself. 23:14.64 Chris Hutchins And i caught a glimpse of a video podcast. it was on a podcast. Somebody's talking about being able to customize your own GPT. I'm like, Hey, wait a second. Did i miss something? I thought I was paying attention. When I realized I could set up my own model inside of that, that platform, i just started dumping my own content in there. So I've, you know I've written and articles and things like that over the years, but just started dumping it into the model. 23:38.59 Chris Hutchins And i had, I'd had a bunch of interview types of things that I had transcribed into it as well. And then I started to ask it questions. And what really was mind blowing to me is it was going back way back, and like 10 years ago, and it was finding something that I might've said or wrote somewhere. 23:59.74 Chris Hutchins And it's bringing it right into the current. and connecting to ah to something that I'm talking about is very, very relevant. Something I'm trying to solve for. It's really, really relevant. I can't remember what I had for breakfast, but chat GPT and then Gen AI went back and found something that was really important well that I had thought about a long time ago, and they brought it back to the forefront. 24:20.70 Chris Hutchins And when I saw that, I got excited first of all. So I go, this is cool. It's going to make me look smart. But what was really interesting to me is realizing that if we can do this for the clinical record using large language models, the physician is looking at a patient that's got a condition. 24:41.21 Chris Hutchins they They've seen it before. My gosh, I can't remember exactly what the details were. I can't remember the patient's name. If I can know the patient's name, I could go find it. 24:52.43 Chris Hutchins And I'm just thinking about these large language models and how this can actually be a game changer to when we're talking about having the most in current information at at the point of care, This can be hugely powerful because even the way you know we use it the prompting now, as primitive as it it might be, it's still a huge leap forward for the physician to have access to do that using large language models. They don't have to be a technical person to do this. 25:22.40 Chris Hutchins You just, frankly, just speak into a microphone and ask the question and then it'll go find something. And it'll tell you exactly who the patient was. It'll tell you all the details that you want to know. 25:32.94 Chris Hutchins And you can you can you do your prompting and inquire until you really get all the the information that you're looking for. and It's not manufacturing it if it's in a language large language model that has actually been developed inside of the firewalls of the house system. 25:48.87 Chris Hutchins but the other The next thing for us too that that I'm excited about, but it's going to be a challenge, is how do we make sure that these large language models don't become isolated so that when you're doing surgery on a patient for the same thing I've done surgery 100 times for, that you have the ability to to get the learnings and the observations from that so that we have a growing knowledge base for that is shared and it's common. the you know The practice of medicine, people get the kind of figure forget about that. 26:23.39 Chris Hutchins Practice means that we haven't solved everything yet and we're continuing to learn. And it's really, really important that we find the ways that make our clinical leaders that are just extraordinary able to benefit from not only understanding and having perfect recall for every case that they've ever seen in their own practice, but what about their colleagues are also in the same practice? 26:48.96 Chris Hutchins What about their colleagues front and in in the national ah societies that they might belong to? it These are all game changers in terms of how much it can improve the accuracy and in the personalized treatments that can be it enabled if we can help them to make this stuff happen. 27:09.72 Chris Hutchins And it's not a small thing. It's going to be a significant lift, but it's one them. That's probably one of the things I'm most excited about from a potential standpoint. 27:20.22 Ratnadeep Bhattacharjee Interesting. Yeah. Chris, just out of curiosity, right? What are some of the other use cases, you know, realistically speaking, that you see getting traction in the next one or two years, especially on the Gen EI and LLM side of things? 27:37.56 Chris Hutchins Well, I think conceptually what i what I just touched on, I know that's happening, um and that that does excite me, but I think the things that are getting of a lot of attention right now and probably has the clinicians more excited is the the use of ambient listening capabilities. 27:57.86 Chris Hutchins And for why I think that makes such a big deal is when you when you're asking the the physician ah only validate what we interpreted from the encounter, right? 28:10.92 Chris Hutchins We're not asking you to trust the diagnosis because that's not what that's not we're trying to achieve. We're just trying to accurately capture the encounter. You don't have to think about it. You're not typing it yourself. At the end of a visit, you can take a quick look and you'll know exactly what happened. 28:26.93 Chris Hutchins If we nail that and we can roll that out at scale, then it makes the next part of the the process really, really easy. um The whole idea of coding, medical coding from that note, that is entirely rule-based. 28:47.00 Chris Hutchins And because it's rule-based, then if we are confident that we've captured accurately that what transpired, what's been documented, then we we can actually and know exactly from that point how to code what it is that they did and they can actually be running against all the, the, this the software that's out there now to, to edit it for validity. You know, you it's wrong diagnosis or, you know, there might've been something transposed. i mean, there's so many different things that can actually be caught immediately. 29:18.73 Chris Hutchins Um, you can say, listen, doctor, you just diagnosed the patient with this. Yeah. Typically, they also have this condition. Have have you have you examined that? Or it might be looking for either the absence or presence of something that should have been identified in the in the visit that maybe wasn't. 29:38.38 Chris Hutchins um The coders actually deal with this stuff. All they do... in their roles, like reading over and over and over again, all the same kind of content and make trying to make sure they're getting the accurate code to be submitted for billing purposes. 29:56.10 Chris Hutchins This job is not an easy one, but just like in this case of where we're trying to eliminate the transcription burden, it also can really make it easier. So a coder is now looking and evaluating things but for exceptions. 30:11.19 Chris Hutchins They're not having to do everything from the ground up. Financial impacts are this are are massive as well. So first of all, you're improving the patient and the provider encounter. Both of them are having a better experience. 30:24.94 Chris Hutchins You're also hopefully not going to make your doctor double and triple their caseload. You're going to actually let them go home and have have dinner, things like that, right? So there's improvement in your provider satisfaction. 30:37.71 Chris Hutchins And then you are going to improve the accuracy of your coding. It's not perfect now, but this will make it more. I mean, there's there's companies out there that are having great success with this stuff already, just in the automatic, the and autonomous coding space. 30:54.00 Chris Hutchins But when you can put these things together, This is going to actually help you to pay for a lot of other cool things because it's going to shorten your revenue cycle time. If you can drop a claim and get it out the door within a few minutes or even in the same day, this goes on for you. know with some Some companies or you do you know that I'm aware of are doing a lot of lu this now. When I was working in Boston, we were doing pretty decent in terms of getting things clean and getting them out the door even you know over a decade ago. 31:22.83 Chris Hutchins But we're talking we're talking about chopping days and days and days out of the revenue cycle time. And it improves the cash flow. It reduces the risk for things being denied. 31:34.42 Chris Hutchins They shouldn't be denied because you're actually capturing what you need to capture up front. So but to me, that that's probably the the place I would say is the next, I don't know, three to three or four years. 31:46.35 Chris Hutchins Imaging obviously is a big one. You know, being able to auto map the images, i mean, that's a game changer. oh But still, there's a there's always that the discomfort in terms of trusting too much. Imaging is very different than reading a document. 32:03.64 Chris Hutchins a lot of the things that you could see in a chest X-ray ah were clearly better indicators of ah the presence of COVID than many other things were um back during the pandemic. I mean, this was some research that I was able to to to look into. 32:20.79 Chris Hutchins And 32:24.83 Chris Hutchins it's again, it's not about... Automating everything. I don't think it's that that's not what we're talking about. It's really adding the, the additional horsepower yeah in the, the processing to analyze as much as you possibly can. 32:40.77 Chris Hutchins Like if you've got 90% of it and it's all mapped, your starting point then is to take a look at the nuance things. But again, you don't want to assume everything because COVID looked like pneumonia. 32:53.67 Chris Hutchins Looks like other things, right? So there's there's a tightrope there you've got to be mindful of. But those are things that I'm most excited about probably in the next like two to three years. I think we'll see some big, big leaps forward. We're seeing some already. 33:09.89 Ratnadeep Bhattacharjee Wow, interesting insights, Chris, honestly. as As you said, right many, many organizations, I feel, are already doing a lot of these, I mean, trying to at least implement these use cases, either through POCs or you know small projects internally, right? 33:28.39 Ratnadeep Bhattacharjee As you can understand, these are all experimentations. 33:32.12 Chris Hutchins Right. 33:32.06 Ratnadeep Bhattacharjee So one of the questions I've always had, and when you were speaking about it, it kept creeping in my mind is that, you know, these experimentations with AI and Gen AI are ongoing, but very few companies actually make it to enterprise level production, right? 33:37.56 Chris Hutchins yeah Right. 33:50.06 Ratnadeep Bhattacharjee What are the key ingredients for scaling beyond pilots, you think? 33:57.56 Chris Hutchins This one is may not be wildly popular to hear, um but I know when you and i first spoke, some of the things that I've been thinking about and and writing about were around data technical debt. 34:13.00 Chris Hutchins um People think of technical debt as an IT problem, and generally speaking it is. um but when you're growing by acquisition, which a lot of the lot of companies are, you're acquiring new systems and their data sets, their workflows, their completions, the the completeness of the data that's in there. 34:36.22 Chris Hutchins um You and I could have gone to the same medical school and we work in the same health system using the same tool. And for perfectly explainable reasons, you're documenting things differently than I do. 34:50.19 Chris Hutchins And that could mean you're putting things in a discrete field so I can easily go get it. And I'm using, I'm just dictating it and it's going in the unstructured note. This complexity and inconsistently it's inconsistency actually can, it exists in real time. 35:08.41 Chris Hutchins and of And even if you've got one integrated health record, this is happening now. So just multiply that by how many physicians that are out there. how many systems they're using, what their preferences are, where they document things, what their training is. There's all these different factors. So there's they's just a really significant difference burden for getting data normalized. 35:36.76 Chris Hutchins And from a thinking about governance, um even three years ago when I was in New York, we had a team of people that are doing their best to move it a good clip to normalize some of the and some of the subject areas so that we can make it easier for you know our specialists, like for endocrinology, for example. 35:56.01 Chris Hutchins We're trying to curate data specifically to answer all the questions that an endocrinologist has to answer and try to eliminate any of the noise that we can. Just standardizing and normalizing the terminology and the naming conventions, understanding that there could be date timestamps that are relevant, depending on which system you're using, which data table things are sitting in, or you might have a data element that lives in several tables and a technologist doesn't know which one's the right one. 36:23.87 Chris Hutchins this this is This kind of thing is really, really problematic. And so if you attach AI to it, well, that's really exciting. But AI is not going to necessarily know what it's looking for. 36:35.74 Chris Hutchins And the mentality that we have got to get past now, and it's a big leap from a trust standpoint, is to be able to use large language models, be able to use the Gen AI to move and and process this and the information faster. 36:52.29 Chris Hutchins Because of the worst and most inaccurate we're going to be is the day we start. Every day from that point, we get better, we get smarter, we get faster. I've have seen things in the last year or so that I realized that this capability is moving now at a clip. 37:08.73 Chris Hutchins where we're talking about hundreds of thousands of objects being normalized in a really, really short period of time. I could have had 10 people working on that manually without that technology, and I wouldn't have gotten a fraction of it done in in three or four times that amount. 37:25.40 Chris Hutchins So we have to be careful that we... Don't hold on to some legacy approach and legacy technologies and start thinking differently. I don't need to, i' not going to be able to ever do what I thought I was trying to do a decade ago. 37:42.86 Chris Hutchins And that is we've got to get everything into a discrete field. We've got to just put some square pegs in round holes. We're going to get people really that are very hard to find by the way. A trained informaticist who has clinical understanding, understands clinical workflow, they understand the the user interface workflows, they understand how the data flows in the back end. 38:01.23 Chris Hutchins That, my friend, is a unicorn. 38:06.70 Chris Hutchins So it's really hard it makes it really difficult to get this kind of work done so you can trust that you're you're going to get the results you're looking for. And so this is, to me, it's hidden beneath the surface and you're always going to have a need for your h your your electronic health record to get attention. 38:13.97 Ratnadeep Bhattacharjee Thank you. 38:23.96 Chris Hutchins It'll need investments. It's going to need people. It's not going to, the the things that cause that, those things to function are nothing to do with what I'm talking about. 38:35.11 Chris Hutchins There's a lot of human factor into it, but there's also differentiators. You build your systems a certain way, um and you believe that there's a perfectly legitimate reason and you can defend it. 38:49.06 Chris Hutchins As wonderful as that is, i still have a problem when, gosh, it never occurred to me when I was designing my system that maybe we'd be absorbed by a larger company one day and we'd have to normalize it. 39:02.52 Chris Hutchins So it's just a problem. it's just It's not easily solved unless we start paying attention now. um If you're early on before you start to pilot, you got to ask those questions. Where are we? 39:15.61 Chris Hutchins yeah How clean is the data? Have we really done our due diligence to be able to know that when we hook this technology to it, it's going to get reliable information. And what has happened in my own experience more far too often is we'll get into something a little bit. 39:31.86 Chris Hutchins This is a development thing. you' I'm sure you've seen it hundreds of times. so You're pretty sure you got a good handle on what you're going to do, but you don't get too far into it And all of a sudden you start so finding some surprises, ah things that we you've expected to be a certain way. And they're really not. 39:47.19 Chris Hutchins ah these are the things that will drag out proof concepts. It'll slow everything down to the point where you're not going to see an ROI in nearly fast enough time to warrant continued investment. 40:01.17 Chris Hutchins And none of these things are inexpensive to begin with. And so it comes becomes very hard for an executive team to defend to their board. When they've made these commitments, they've invested all this money, 40:14.82 Chris Hutchins where's where's the result? And when do we scale this thing? it's I wish there was an easier way to to talk about it, but this is something that's just been hidden for such a long time. My budgets were always buried in an IT budget, for example. 40:30.56 Chris Hutchins And I'm like, I knew I'm not going to go get any really cool software this year because we have some aging infrastructure. we've We've got some systems that are really inadequate and we're going to have to do a rip and replace. 40:44.07 Chris Hutchins ah We're actually acquiring more systems than we're going to be needing to integrate that data. it's a problem that can perpetuate itself and usually does. But as you're starting, you know, to, to consider which things you're going to test from piloting standpoint, these are important things to look at. What is our data readiness? 41:04.03 Chris Hutchins Is it normalized? Can we trust it? And i don't mean asking the it people or the data people, can you trust it? ah We're going to have we have to get some clinical leadership that is a, 41:16.74 Chris Hutchins excited about helping us to do the validation. And and again, there's the unicorn problem we have. 41:24.65 Chris Hutchins But these are things that are really, really important to to think about before you start down the road of of defining the pilots you want to do. 41:34.36 Ratnadeep Bhattacharjee Interesting. No, uh, as he, as he kept saying about why, um about the why of, uh, all these, uh, pilots or small projects failing or rather not giving the ROI that. 41:49.79 Ratnadeep Bhattacharjee Potentially it should have, right. Uh, one of the things that I wanted to understand was, let's say not at this moment, but let's say if we fast forward five years. 41:52.50 Chris Hutchins Right. 42:01.59 Ratnadeep Bhattacharjee What does good look like when it comes to AI-enabled healthcare that's truly patient-centered? 42:16.92 Chris Hutchins like The obvious answer for me is it's it's got to be better than it is now. but and i mean that's yeah i'm i'm not going to bail on you that that quickly. 42:28.21 Chris Hutchins But I think what we're if we're if we're paying attention, we have to, as technologists, investors, leaders, get a little bit more in tune with how they how our clinicians think and how they operate. 42:46.83 Chris Hutchins There's risk always, there's and there's nothing we can do about that. But it really has to start with that patient and provider encounter, focusing on where the friction points are. 42:59.81 Chris Hutchins Things like simply simple components like how scheduling and getting access to see your physician. These are things that if if we don't do some really basic things very well, 43:13.95 Chris Hutchins This is going to continue to be a problem. And so we just have to start thinking a little bit differently. Let's not worry about the cool technology for a second. um To my knowledge, I had some teams that have done some really cool stuff with data. Their dashboard designs, the integrations, unbelievable. 43:31.64 Chris Hutchins But not one person ever walked through the front door of a doctor's office because we had cool tech. They didn't know anything about our our dashboards. they don't They don't need to know about that. 43:43.63 Chris Hutchins It's like, if you can see the sound guy when you go to see your favorite concert, your favorite band, it's not good if everyone knows who the guy is because he's visible. 43:56.39 Chris Hutchins We have to figure out how do we do the things that are just eliminating the burdensome aspects of what it takes to to deliver that frictionless encounter for the patient. 43:56.71 Ratnadeep Bhattacharjee Absolutely. 44:10.95 Chris Hutchins And it starts with the patient. It ends with the patient. But it also, I'll back it up a little bit from an ah higher level view. It's people. Right? It's the physician, it's the nurse, it's all the support staff, and it's the patient. 44:27.13 Chris Hutchins Everything has to be wrapped around that. If it's not about people, ah we need to pause and rethink what we're doing. Because this, at the end of the day, is only going to be meaningful if it's going to help a person. 44:41.25 Ratnadeep Bhattacharjee Right, right. No, I agree, ah Chris. you know Before we wrap up, Chris, what advice would you give to ah the digital health founders and enterprise leaders who want to deploy Gen AI or large language models within their enterprise you know the right way? 45:07.81 Chris Hutchins I think just kind of going back to what i was just touching on a little bit, let's start with finding out what the pain points are. What are the problems that would that really need to be solved? 45:18.82 Chris Hutchins And understand that healthcare care has pressure unbelievable pressures from a financial standpoint, never more so than they do now. um The dynamic that we're seeing in the political arena um shouldn't be surprising to us because we've kind of kicked the can down the road for over a decade now. 45:34.28 Chris Hutchins on things that we knew we had to find a way to address at some point, the cost keeps going up, right? And now the big pharma is getting a lot of attention. So I don't think the dynamic that we find ourselves in is going to be changing dramatically. And you know all of a sudden we're going to be able to invest in things and it's going to be great on the provider side in particular. 45:55.69 Chris Hutchins It's really not going to be that way. So we if we're thinking about How do we go understand what the pain points are and make sure that we're thinking about solutioning for them? 46:06.88 Chris Hutchins But the partnering component of it is going to be a really important aspect of it. We need to have the the trust and the confidence of some clinical leadership within organizations because they will advocate for things that If they know and they believe that it's going to make the patient encounter better, it's going to improve their ability to deliver better care. 46:29.83 Chris Hutchins That's what they care about. So I think that's probably one of the first things that I would say we have to think about. First, what is the real problem that that's that's existing? We don't have to assume you could probably go into almost any physician's office or hospital and just pick a random one, ask them what their pain points are. 46:49.74 Chris Hutchins They can probably give you a whole bunch of them. And those are the types of things that we want to be able to start with. And, you know, for me, I think how you and i ended up here today is just thinking about how do we come together and solve for this? Because it's it's going to be a combination of investors who have a passion to do something good with with their resources. 47:10.29 Chris Hutchins You have to pair that with the technologists that are like really excited about applying their technologies to solving these problems. You get those two people coming together And now we get the partnership on the other side, the ones that who who took the Hippocratic Oath. 47:26.03 Chris Hutchins We kind of come together in ah and a team kind of ah approach to this where we're all here. First, do no harm. It's really that simple from our starting point. 47:37.67 Chris Hutchins then it's about how do we improve the lives of human beings? Because after all, it might be you or me. It might be one of our family members. It could be our neighbor. That is going to depend on us coming together and doing this because it's the right thing. 47:51.20 Chris Hutchins And if we can do this at scale here in the U.S., the way that i know that it's possible, it's going to change things globally. And I feel like as a citizen of the of my own town, I'm I owe it. 48:06.18 Chris Hutchins But as citizens of the world, don't we all? Right? And if if we're not thinking big, then we're we're not going to challenge ourselves nearly enough. 48:17.87 Ratnadeep Bhattacharjee I agree. I agree, Chris. You know, this has been lovely. Chris, this was really and incredibly insightful. Thank you for taking the time to share your experiences and perspectives. 48:31.36 Ratnadeep Bhattacharjee It's been a pleasure, honestly, having you on Leaders Perspective. ah And to our listeners, if you found value in today's episode, be sure to check our past episodes and subscribe more for conversations at the intersection of data, innovation, and healthcare leadership. 48:48.24 Ratnadeep Bhattacharjee Thanks for tuning in.
