00:00.96 Ratnadeep Bhattacharjee Hello and welcome to another episode of Leaders Perspective, where we talk to healthcare and technology leaders who are shaping the future of care delivery and innovation. 00:02.14 shannon kennedy Thank 00:10.43 Ratnadeep Bhattacharjee I'm your host Ratnathipaticharji, co-founder at TechSagyabit. Today's topic is laying the groundwork for AI-ready healthcare, where we will discuss the essential steps healthcare organizations must take to prepare their data and infrastructure for their AI-driven future. 00:27.25 Ratnadeep Bhattacharjee We'll explore how standards like HL7 and FHIR, data quality, interoperability, governance, all all of these form the backbone of meaningful AI adoption. Joining me today is Shannon, a healthcare care strategy and technology advisor at SETMED Advisors and Product Edge. 00:44.91 Ratnadeep Bhattacharjee At Sekhmet Advisors, she partners with healthcare organizations to create strategic roadmaps for interoperability, AI readiness, and digital transformation. At Product Edge, she advises on building and scaling healthcare products that leverage standards, compliance, and innovation to meet the demands of the rapidly evolving digital health ecosystem. 01:03.04 Ratnadeep Bhattacharjee With her expertise at the intersection of strategy, standards, and data-driven innovation, Shannon brings a unique perspective unique ability to bridge technical and business priorities in healthcare. 01:14.34 Ratnadeep Bhattacharjee Shannon, welcome to Leaders Perspective. It's wonderful to have you here. 01:18.48 shannon kennedy Thank you so much. It's a pleasure to be here. 01:23.61 Ratnadeep Bhattacharjee Shannon, to start us off, you know, why do you think is data readiness for AI, you know, with with all the buzz going around in terms of AI, right? Why is data readiness for AI such a critical conversation in healthcare care today? 01:38.44 Ratnadeep Bhattacharjee I mean, what gaps do you see most organizations struggling with, right? When they when they think about implementing AI solutions? 01:49.03 shannon kennedy Yeah, so i'll I'll probably start at a little higher level because it's usually the humans that require some data understanding before the AI can even compute it. So i i try to start at that level. 02:04.31 shannon kennedy I think that the gaps I'm seeing right now, frankly, are not much different than the gaps I saw 20 ago when ah Cloud was starting to come out when we were starting to even see a lot of regulations in the US push interoperability, push electronic health records. 02:24.36 shannon kennedy and And really what I see as the biggest barrier for humans is risk ah tolerance. So a lot of fear. And so at the very top level, 02:37.93 shannon kennedy you have this spectrum, I believe, of fear in boardrooms and at executive levels because fundamentally people don't understand for the most part how AI works. 02:51.49 shannon kennedy And so for those of us who have been doing any kind of applied math for a while or any kind of statistics or any kind of data science, we know that this is really old math still for the most part And it's pretty straightforward ah formulas or computational methods. 03:14.13 shannon kennedy But the the power of the compute has just become so much more, right? So much more powerful. And so the ability now for the math to be done in these multidimensional levels is is pretty incredible. 03:32.72 shannon kennedy Are you still able to hear hear me? Because I have a message that says I'm offline. 03:42.53 shannon kennedy oh I can't hear you now. You're on mute, it looks like. 03:48.18 shannon kennedy OK, my apologies. 03:48.47 Ratnadeep Bhattacharjee No, no, you can continue 03:50.22 shannon kennedy ah So once you can kind of overcome some of the understanding around AI, I think the next level of understanding is, you know, where are we with this technology? 04:05.96 shannon kennedy And I think it's still very early days, right? I think it's still very early days. I think again, those of us who have been following this for a number of years, you know, AI has been around for a very long time and so has data messiness as I'll call it. 04:23.09 shannon kennedy And I think in healthcare, care there's an even more unique challenge because as you know, being data ready in a industry that still uses fax machines and still uses paper records is, 04:40.74 shannon kennedy you know, it's comical, right? It really is. And it's, it's, it's one of the reasons I, I honestly feel very frustrated about healthcare in general. 04:52.97 shannon kennedy And it's one of the reasons why I ended up, you know, really leaving industry to start exploring, what can I do um as a member of technology teams? Where can I start to work with people who are really laser focused on solving this problem? 05:11.03 shannon kennedy um and And to be, again, completely, you know, fair and and also a little bit cynical, my drive and my passion really comes from the perspective of the person. 05:24.55 shannon kennedy And when I say person, i mean the patient or I mean the member or i mean the human who hasn't even received their health care yet. And so I definitely feel like that is said a lot. 05:39.81 shannon kennedy in healthcare care that, you know, we're people centered or we're patient centered. But when it comes to technology, it really starts to be more about its revenue centered, its billing centered, its, you know, workflow centered, it becomes less about the human and more about the very kind of short term needs of of driving revenue, which I understand that part too. But What gets me so excited about AI is I think this is going to be the technology that helps marry both. 06:12.43 shannon kennedy So back to your original question about data and data readiness, I think that, you know, a couple of things, um depending on the size of the organization, ah the smaller you are, i believe the more of an edge you have. 06:27.74 shannon kennedy I believe that you cannot be too small in health care. to start taking advantage of even these early days of AI. Even if it's AI to get your data ready, if it's a matter of trying to figure out just where to start, you can literally start talking to Claude or a ChatGPT and start educating yourself. 06:52.76 shannon kennedy And ah what's so funny to me is, you know, I talk to i talked to my parents who are nearing their eighties all the time about this topic and they're getting more and more engaged. But originally when I started talking with them, they're like, well, i don't know what you mean about, you know, you know, data and this AI and this healthcare stuff. What does this even mean? 07:12.97 shannon kennedy We don't know how to use AI. And and then literally my mom says that. And two minutes later, she's asking Siri to set the timer so that she knows when her chicken's going to be done in the oven. 07:24.99 shannon kennedy And I said, mom, you have the ability now to literally talk to chat GPT, just like you talk to Siri and you can start asking her anything. And so once I got them turned on to that, you know, my dad's asking all kinds of statistics, it's like encyclopedia on demand. Right. So then the, you know, back to that human element, starting to get familiar with it being a tool, being co-pilot. 07:48.14 shannon kennedy And I think that the smaller you are as an organization, you can really start utilizing it just at that base level. You know, even with the health care at the human level, ah another great example, ahll you'll hear me use my parents a lot because um all my kids are out of the house now and I live just two doors down from my parents. So I have this awesome compound that I get to go back and forth with. 08:11.64 shannon kennedy But, um you know, my dad has had colonoscopies historically every five years for a number of years because he has polyps. I'm sure he won't mind that I share that about him. 08:22.94 shannon kennedy because this is so funny. And for years, he has not been able to understand his results. He has no idea what those pictures mean. This is one of the only professions besides you know law that still is using Latin to communicate in medical records. and And so when you go and you look at your lab results or you look at a test or a radiology report, people are lost. 08:49.36 shannon kennedy They don't even understand. And so I told my dad, I said, dad, just take a picture of those results and leave your name out of it. And then ask, you know, ask chat to tell you what it means. 09:04.22 shannon kennedy An hour later, my dad is calling me saying, you need to come over here. This is what's going on. You know, I have this and this and this. And I want to show you this picture because this is where it's at. And I went over there for dinner. And literally because of AI, my dad was able to understand his colonoscopy. report And so, you know, you'll I keep going back to kind of this human-centered or this person-centered thing, but I really believe that when it comes to like data, data readiness, just fundamentally understanding how can AI be a tool and how can it really democratize knowledge. 09:38.44 shannon kennedy um And then on a little bit more fundamental level, which is, you know, what all of my technology colleagues are thinking about, I know i call myself a technology strategist. 09:50.53 shannon kennedy I am certainly not an engineer. i know enough to be dangerous. i'm I'm the one that translates, well, this is how the humans do the work. And this is where the data lives. um And this is the three pieces of data that you need. 10:04.70 shannon kennedy When it comes to to data readiness, the other thing that I see with with people is they just, they see this huge elephant of effort. And in reality, 10:15.70 shannon kennedy You don't need every piece of data. When people look at medical records, for example, for those of us who have been looking at medical records for a very long time, especially those of us who have been auditing medical records for a very long time, there is a ton of information in medical records that is just plain waste and bloat. 10:38.86 shannon kennedy So if you think about when you go to the doctor and they give you your papers, you know first of all, even if you can understand them, but if you were going to take pictures of all of them and have even your chat GPT read it to you, you know, there's probably 15 pages that you're going to have to scan, but maybe one of them has something relevant in it. 10:57.42 shannon kennedy And that's, you know, a lot of that is regulatory driven, a lot of notices, a lot of this, a lot of that, a lot of instructions, a lot of education, another thing I could talk for hours on. But um when it comes to data readiness as well, I think that people get very um overwhelmed and they think, oh, my gosh, I have to have everything, ah you know, data ready. And that's just not the case. 11:24.38 Ratnadeep Bhattacharjee Right, right. No, this is this is a very fresh perspective Shannon, you know Putting people at the center of whatever is happening in terms of you know You can be data ready you can be air ready, but if there is no empathy like that, it doesn't work. 11:40.03 Ratnadeep Bhattacharjee but Let's put it down now ah So to shift our focus a little bit on on a different aspect of data readiness, right? You know, nowadays standards like HL7, FHIR are often kind of seen as compliance checkboxes rather than really strategic enables, right? 11:59.58 Ratnadeep Bhattacharjee So from your experience, Shannon, how should organizations view these standards in the context of probably long-term interoperability and eventually, you know, AI success, right? 12:10.05 shannon kennedy Yeah, I go back and forth on this topic because, you know, when we talk about data standards, again, what folks need to understand is, you know, you're you're trying to figure out how do I build a pipeline, right, that it allows the the data from my systems to be siphoned out and then computable. 12:37.79 shannon kennedy Or how do I build a pipeline to then be able to interface and communicate with another system? And again, this is where, when I think about those standards, and and I've been around a long time, so I've i've participated in a lot of the early experimentation. I've been very plugged in with both HL7 and FHIR, and very, very early on participating with some large academic medical centers through some of the the original DaVinci and even last year I was a member of Sequoia. So there was um there's been a lot of ah um effort on that front. 13:20.03 shannon kennedy But even at that level, i often wonder if we're going to get to a point where some of that patchwork becomes irrelevant. I'm very excited about where AI is going to be able to take data readiness, frankly. 13:36.55 shannon kennedy I imagine a world where there is going to be a Gentic AI setup that is able to essentially go and sleuth the data and create some of those pipelines on our behalf until a particular system is um eventually you know shelved because it's it's now either served its use or we're working in a completely different environment where that type of duct tape or patchwork isn't needed. 14:08.76 shannon kennedy So I'm already probably decades ahead. But in the meantime, I do believe that with most of the projects, at least that I'm dealing with, being able to identify what are the endpoints that we actually need, understanding that we do need to have some kind of a common language at this point, 14:27.98 shannon kennedy You know, a lot of people who have been in this world for a while, I mean, HL7 is iss it's still very, very old. you know, this is not new stuff. This is old, old, old interface methodology. 14:41.67 shannon kennedy And, you know, FHIR is just another standard that makes it a little quicker. And so I think it's, you know, it's been an iterative approach. which is great because you do need to have some common languages. The way I explain it to folks is, you know, imagine if you had this one computer system and it speaks French and then you have another computer system and it speaks Japanese. 15:04.58 shannon kennedy Well, you need to be able to create an interpreter between it. And the only way to do that is have the interpreter essentially be able to plug into the French machine and ah be able to translate it to Japanese so that when it gets to your machine, 15:19.93 shannon kennedy that you understand it. And so that's a very kind of simplistic way to explain it. But I do think that um people also get very, very confused and overwhelmed of what is even an API, right? I do think that for those of us that have been in this world being API ready or even building out technology anymore that, it i mean, does anybody even build technology anymore that's not API ready? That just seems like crazy. 15:45.14 shannon kennedy But there are a lot of legacy systems out there that are not API ready, that are not ah using any of these these types of of interfaces. So they sit in the siloed these siloed databases. 16:01.08 shannon kennedy The other thing too that I see with data and even the ability to plug into them is the the code itself. That seems like it's all just hanging on by a thread. 16:13.25 shannon kennedy right? It's like been developed upon and developed upon that if you even try to pipe into it, it potentially disrupts an operation that it's, you know, living and breathing. And again, this is where you can see, i get very, you know, less technology and more of how does it get dumbed down for me. 16:32.83 shannon kennedy But oftentimes, I'll have my engineers tell me, well, we can't actually go mess with that, because that's, it so it's an operational environment for them. And if we even go try to pipe in, it's going to completely you know screw up the duct tape they have on that particular workflow. 16:48.23 shannon kennedy Right. And so that to me also is is where I see, even if you have you know this these endpoint readiness, or even if you're if you're trying to implement any kind of interface, whether it's HL7 or even make it quicker with FHIR, 17:04.40 shannon kennedy Amy Nunez- Those standards if you have a system, a legacy system that's still know paper clipped and duct tape together then you're still going to have trouble. 17:19.62 Ratnadeep Bhattacharjee no That's fair. you know Just related to this question, like ah have you have you in in the and in within the industry itself, have you encountered examples where these standards early on has helped organizations scale AI or are helping organizations kind of get to AI or even scale AI and analytics initiatives faster? 17:43.22 shannon kennedy Yeah, that's, I mean, just to not to be too much of a plug, but I think that, you know, your organization does that, Productive Edge does that. And I think that that is going to be an amazing, that kind of the next iteration of data readiness, where you can come in and you can essentially not have to rip and replace, right, a whole system. 18:07.00 shannon kennedy But like we say at Productive Edge, you know, supercharge your current systems or supercharge your current humans and be able to actually use AI to bypass some of those current systems. So for example, again, if we imagine, you know, we have four systems and, you know, they aren't talking to each other, or maybe they are, and we've built these, you know, very delicate interfaces ah that, you know, are, who knows how, how reliable they are. 18:40.68 shannon kennedy ah The nice thing about, in my opinion, AI now is that you can sort of create these pipelines and you can create a separate platform that's cloud-based or on-prem, whatever they want to do. 18:55.38 shannon kennedy And you can start to kind of suck the information using you know and and an API, whether it's an HL7 or FHIR-ready API, and being able to kind of suck the information that you need to now aggregate some of that information from those multiple systems and and create an environment where now you're basically pulling all of that together and now you can interrogate it. 19:25.32 shannon kennedy That is something that is very, very valuable for folks who are working out of multiple systems and production environments, whether it's call centers or it's revenue cycle or it's any kind of of work that people are doing where they're having to either use radio buttons or have to go into multiple environments to do their work. 19:50.08 shannon kennedy So i'm definitely seeing a lot being done on that front. 20:02.81 Ratnadeep Bhattacharjee Yeah, so Shannon, I also wanted to ask you about some of the pitfalls, right, you know, especially on data quality and covered inside of the city, even with the right standards, 20:18.12 Ratnadeep Bhattacharjee you know, poor data quality and you know poor governance often often derail AI projects, right? what According to you and as per your experience, what are some of the most common pitfalls that organizations face, right? 20:30.91 Ratnadeep Bhattacharjee And how can they proactively, you know, address these issues? 20:34.39 shannon kennedy Yeah, you know, it's interesting. I think that, ah you know, the saying garbage in garbage out, right, we hear that all the time. um i would say that some of the the pitfalls that are the most common that I see is pulling information from a system that is not a source system, and then something gets overwritten. 20:58.19 shannon kennedy um or looking for data where there's a lot of what I call inter-rater reliability. So humans do the work differently. 21:09.58 shannon kennedy um Other pitfalls I see are the workflows are done differently. So the data fields where the data is intended to go is not going in that particular place. 21:21.06 shannon kennedy So for example, in healthcare, in electronic health records, there is a lot of box checking, lot of clicking, right? I, um I remember again, in the very early days, helping to implement these huge electronic health records that everybody was so excited and so optimistic about. 21:43.06 shannon kennedy And I think if you ask anyone today, you know, where we've come, most people would tell you, i kind of wish we were just, you know, back to transcription because now I feel like how people, 21:57.40 shannon kennedy would rather be able to do their work, especially if you're patient facing is to have a sensor on and to be able to have a ah conversation with the the patient and to be able to have that information fed into ah system that can then aggregate and assign and push through to the workflow that it's supposed to um Because right now, if you look at how people do their work, 22:28.07 shannon kennedy oftentimes they have an option. They can check a box or they can use a free text and write a note. And so the biggest issues i see with with with data ah pollution, as I'll call it, is folks do not use systems the same. 22:46.43 shannon kennedy this is This is human driven work ultimately that is deriving this data. Unless you're in a sector of healthcare that truly is binary. 22:59.71 shannon kennedy So lab values perhaps, or other types of systems where you don't have a lot of of notes or interpretation um using words, although there's a lot you can do with NLP these days. 23:15.73 shannon kennedy That's where I really see a lot of the challenges is that there's assumptions made about where data lives. And then when you actually go and you watch where people are doing their work, I call it, you know, 23:26.11 shannon kennedy Let's go watch and see what's happening in the wild. Right. I have s SOPs. I have business analysts telling me what's being done, but I never believe it until I actually go and I watch the work being done. 23:40.94 shannon kennedy That's that's my world. I understand operations. you know, I understand how to see how the humans are doing it. I understand how humans are trying to get as much done as possible. So there's always going to be some kind of a workflow hack. 23:55.99 shannon kennedy And when we don't anticipate or we're unrealistic or we make assumptions, even as technologists, that this is how the work is being done because this is what the business analyst is telling me or this is what the, you know, the spec says or the documentation says or, you know, this is this is what the SOP says. 24:14.15 shannon kennedy I don't believe it until I see the work being done in the wild. Once I see the work being done in the wild, usually. I can identify that we're going to have inter-rater reliability issues because people do the work differently and the data goes in different places. 24:28.14 shannon kennedy Or we don't um we're not really even following that standard. The data is actually getting inputted here. People aren't using the checkboxes because it's easier for them to just write a note. 24:42.75 Ratnadeep Bhattacharjee Right, right. No, no, this is, this is, I agree with you, right? But Shannon, oh moving on to, you know, better things or, you know, moving on to the future of AI, right? 24:55.63 Ratnadeep Bhattacharjee So as, as you know, this Gen AI and agentic AI systems enter the healthcare space, how will the definition of AI readiness evolve to it? 25:06.87 shannon kennedy Yeah, that's that's a great question. i think that when I think about ai readiness, especially in healthcare, care you know, I do think you have to take it one kind of business function or business unit at a time. 25:29.25 shannon kennedy The ability for firms like Productive Edge or your firm or or consulting to come in and be able to help with governance, help with an assessment of where your data lives, what do you want to even get started? 25:49.11 shannon kennedy i think it has to start at that level ah from a data readiness or an AI readiness perspective. I'm a firm be believer in trying to incubate good ideas from your own workforce as well. 26:05.62 shannon kennedy especially if you have operators who are also technology savvy. I used to be one of those folks. I was one of those folks that would be, you know, in operations and i would have, I would be doing a lot of experimentation with my teams around technology, or I would be enabling other teams to do a lot of experimentation. 26:26.84 shannon kennedy But oftentimes a lot of those really good ideas and those, those innovative, um kind of workflows, they get they get squashed because there the rest of the organization, the virus, as I call it, the virus comes and and you know it takes over you know whatever the germ is of innovation. 26:48.89 shannon kennedy And so I think that in order for folks to be AI ready, they have to first look at change management and they have to understand what is their culture for readiness. 27:00.59 shannon kennedy If you do not have an innovative, forward-thinking leadership team, AI readiness is never going to come. It won't. And i think that it's go to a lot's going to tease itself out over the next probably, gosh, I would normally say 10 years, but I think it's going to be shorter than that. I think that within the next three, at ah maximum five, we're going to start to see organizations pull ahead that are forward-thinking and innovative thinking. 27:33.44 shannon kennedy And more often than not, you know, what we're exposed to when it comes to innovation is what people see at a conference or, you know, a fancy article or what they see on social media. 27:44.98 shannon kennedy But when you get into the wilds, when you actually get into the healthcare organizations, for the most part, it's still being done the way it's being done. And so I think that emerging nations and emerging economies are actually going to be AI ready before the West. 28:05.49 shannon kennedy I think that if you start to look at what's happening in India, in Africa, and South America, in countries that are not bogged down by this, you know, sunk cost, they're going to be the ones that I feel very optimistic about leading the way. Even China, if you think about ah the the nations that the West competes with, 28:32.53 shannon kennedy they are AI ready from a mindset, from a ah a desire, at desperation and need. you know, that it's the folks that are are being the most innovative are not the ones with the resources, it's the ones with the need. 28:48.19 shannon kennedy And um I think that the West really needs to figure out if we're gonna be AI ready at an organizational level, then then The organizations need to start educating themselves. 29:01.75 shannon kennedy They need to start getting uncomfortable. They need to start incubating with the bright minds in their organizations that have good ideas and lifting those to the top. And they need to start bringing in experts to assess their environment and find out what is our AI strategy. 29:17.69 shannon kennedy Don't just promote the, you know, the CIO and say, guess what? You get to be the chief AI officer now too. As you and I both know, being a technology person, 29:29.65 shannon kennedy is not the same as being an AI expert. You'd be better off hiring a data scientist straight out of one of the great labs, you know, to to bring them in to help you before you look to your technology team. 29:45.55 shannon kennedy If anything, I think that they're even more afraid of what AI might become. 29:53.79 Ratnadeep Bhattacharjee Right, right. Shannon, ah you know, this has been a deeply insightful conversation, honestly. know Thank you for you know breaking down what it really takes to be AI ready generally. You know, it's, I mean, in ah in a very non-technical way, in a very human way. ah So before we wrap up, ah Shannon, where can our listeners kind of follow your work? 30:19.60 shannon kennedy Absolutely. Yeah. So I spend most of my time with Productive Edge. And so you can come to our website at ProductiveEdge.com. We are a team of technologists and operators and former leaders in healthcare are really trying to change change the world. 30:37.39 shannon kennedy I'm very proud of our work. And then you can also find me at The Digital Economist, which is a global think tank that is also trying very hard to change technology for the world to make sure that it is human centered. 30:55.65 Ratnadeep Bhattacharjee ah Thank you again, Shannon. And to all our listeners, thank you for tuning in. If you found this episode valuable and would want and know Shannon to come for a part two, please subscribe and share it with your peers who are passionate about healthcare innovation. 31:13.08 Ratnadeep Bhattacharjee This is Ratnati Bhattacharji and you have been listening to Leaders Perspective. Until next time, stay curious and stay innovative.
