Empowering Clinicians Through AI and Data Innovation with Spriha Gogia of Ophelia

In this episode, Spriha Gogia, Senior Director of Data at Ophelia, shares how embracing data chaos and aligning analytics with business strategy can transform healthcare operations. She also breaks down the evolving role of AI and its impact on patient care.

‍

Key Takeaways:

‍

00:00 Introduction.

02:42 Spriha’s passion for science led her from academia to data-driven healthcare.

06:46 Healthcare data spans systems to wearables, and data teams must make it cohesive.

10:14 Data silos persist, but an enterprise data warehouse brings order to chaos.

16:32 Data leaders should embed in strategy discussions to align with business goals.

20:08 A data product is any analysis or dataset that delivers value to its stakeholders.

24:28 AI mimics human tasks — machine learning predicts outcomes autonomously.

30:10 GenAI can ease clinician burnout by automating repetitive documentation tasks.

35:25 OUD care generates persistent, complex data requiring ongoing patient tracking.

‍

‍

Spriha: [00:00:00] data can do a lot, and the biggest and most important pieces of the business is where the, their team should try and focus, and become a proactive force rather than, you know, always being reactive to needs.

‍

Anthony: welcome to the Data Masters Podcast. Our guest today is Riha Goia, the senior director of Data at Ophelia, and recently named one of the 100 most influential AI leaders in the us. ESP has a fascinating background having transitioned from a PhD in biology and a master's in [00:01:00] public health from Columbia to becoming a leader in data analytics.

‍

Today, we'll explore her career journey and specifically her specialty of building data teams from the ground up at organizations like NYC Health and Hospitals, and now Ophelia. And I'm interested to get her expert opinion on two of the most ambiguous terms in our industry. Data, products and ai. Finally, we'll dive into the role that AI is playing in healthcare and get her perspective on how to empower clinicians and improve patient care without losing an essential human element.

‍

Spriha, welcome to the show.

‍

Spriha: Thanks for having me. It's an absolute pleasure to be here, and to talk about my experiences working in data. say that while I'm excited to talk to you today, everything I say here is my personal opinion and may or may not reflect the stance. Ophelia has a. For that matter, any of the other organizations that I've worked at.

‍

Anthony: Of course. so [00:02:00] let's talk a little bit about your background and career. 'cause I think it'll help, level and frame the conversation as as we dig into it. you've had a fascinating career. you started, in a PhD in biology and moved into public health and data and analytics, and a recurring theme.

‍

as we prepared for this conversation was this idea of building data teams from the ground up, either in something like a public hospital system or a startup, and of course now at Ophelia. so maybe share a little bit more about your background and then we can dig into, kind of what you've learned from that experience, but maybe help people with a little bit of grounding.

‍

Spriha: Yeah, absolutely. I'm happy to. so I think of myself as a science nerd. I was always, into, you know, learning more about, nature in general. And that led me to, you know, really take up a master's and a PhD in biology. and I spent, six plus years, getting, in academics. and while I loved the science part [00:03:00] of it, I did realize that I, I wanted to be more, , social and not be away in a lab essentially. and that, made me realize that I wanted to be in the industry and, you know, be talking to people. that brought me to public health. and I've been in healthcare data analytics. Ever since. and it's been a great journey, honestly. I love the impact of, you know, the science that I do. I still do some science.

‍

There is a lot of data everywhere. and science is essentially data. but I see where the data is taking, you know, having an impact and that's what excites me.

‍

Anthony: So, obviously your focus has been on data and in healthcare and, and you've spent a lot of energy thinking about how data can be help. On healthcare, I often make this joke that, the US government, or governments in general, I suppose, and healthcare are probably the two areas that have the messiest and most complicated, most siloed, data.

‍

I don't know if you would agree with that statement and, and what your experience of data in the healthcare industry specifically, what's made that? So [00:04:00] is that different, you know, unique.

‍

Spriha: well first I, I know this is a Data Masters podcast, but I should preface this by saying that data is everywhere and in everything that we do today, right? Essentially, in today's day and age, it is hard for us to accomplish anything without interacting with some kind of technology. And every technology then generates data.

‍

So think about any interaction you have. you are generating data. So now getting to your question about like, is there complexity and massine cent data? Absolutely. That's true across the board, and especially true in healthcare as it turns out. so I will, you know, walk through an example of what healthcare data looks like, and hopefully like that will illustrate how complex it can get. so think about, you know, healthcare interaction where you're going to see your doctor. and let's think about the various places where we are generating data in that. So first of all, we probably need to schedule an appointment. [00:05:00] So that's like a scheduling data set, right? And turns out that already it's fairly complicated. working at a healthcare organization. There are so many clinicians and so many calendars, and we are all always trying to sort of like. Figure out and fit in, you know, multiple appointments of different time periods and different schedules and all of that. and I, so far I've only talked about clinicians.

‍

Of course, there are many different role types, to consider as well. Okay. So from there onwards, even before you've gone to your appointment, the healthcare organization is now trying to organize your care, not just in terms of the scheduling, but also. Your, insurance, right? So, they will also think about like all of your data that you have from previous visits, if you've been with them before. and all of this is even before you've made it to the doctor's appointment. So we've now talked about types of data. So there's the insurance and the financials data, and then there's like the preclinical work before. Even making it to the visit. And in some cases after you make it with, or in most cases, I would [00:06:00] say after you make it to the visit, there is somebody who's evaluating you at the front desk or the nurse is talking to you. and all of this is data being generated and documented. then now we are finally in the clinical encounter where the patient is talking to the clinician and the clinician is spending a lot of time, you know, taking notes and, talking about the diagnoses or the symptoms that the patient might have or procedures. and then after the visit is completed, there is data around, you know. Actual getting billed or generating the revenue behind all of this, these transactions. if the clinician prescribed any meds, then there's data around the pharmacies. and there's a lot of other types of data to consider as well.

‍

So if you are at a like medium or a larger size healthcare organization, you probably have a variety of employees that you care about. Maybe you have more than one. Organization or, sorry, physical location to think about. maybe you have, you wanna think about like the number of beds you [00:07:00] have. so all of this is essential, in considering when you are a healthcare organization. there are newer types of data to consider as well. So, think about the health apps we now have in all of our phones and smart watches. They are constantly recording data about our lives. How long did we sleep? how many steps did we take in a day? and based on, you know, like what you want to talk to your provider about, that could be data for the clinician to be interested in as well. so. just talked through like all of these types of data, and it turns out that it's the job of the data team or the data leader at an organization to pass through and figure out, Hey, what parts do I need to be able to tell a meaningful and cohesive story about what's going on at that organization?

‍

Anthony: So I, I, I think in a funny way, you've. Properly summarize all of the biggest challenges in data today. So just to [00:08:00] to reiterate like there's a silos problem because your interaction with a doctor, the Dr. May or may not, in a sense work for the hospital. They may be independent, they may be there. They certainly don't work for the insurance company.

‍

The insurance company's a completely separate entity. Forget about, if you're getting, blood work or something, you know, tests that are separate. Those are often completely different companies. So we've naturally built in a siloed data environment. Then you also mentioned the, and something that's often come up in the podcast, structured versus unstructured data.

‍

So there's a lot of, you know, the good part about a schedule as complex as it may be, at least it's well structured. in contrast, the doctor's notes are, either illegible and certainly not. structured there. They're just text. but you also mentioned that I would agree iot data can be both very valuable, but also, again, a wildly different stream of data.

‍

And last but not least, there's the obvious question around security and privacy. So obviously much of this data is not something that people would want spread [00:09:00] around. And even within a set of data, what the doctor knows versus the insurance company, need to be well, segmented. It feels like. the challenge in the healthcare space brings all of the most complex questions around data into one place.

‍

Is that fair?

‍

Spriha: Absolutely. That's one of the bigger challenges I would say. and honestly, if you ask me what do we do about it? I think we have to own the chaos, you know, because it just exists. And so, then the question is like, how do I deal with it? It's really like, what is the problem that you're trying to solve?

‍

Anthony: And then let's try to solve the chaos in there because we absolutely cannot make sense of everything because there's always newer types of data that's coming in. and that's actually good. So given the level of complexity and, and challenges, and based on the introduction, you've spent a lot of time building data teams and often from scratch. so in a good way, like you've, you've worked to [00:10:00] solving this problem. multiple times at different kinds of organizations, from startups to large healthcare organizations.

‍

are there some key lessons learned that you've taken from those experiences that others could benefit from?

‍

Spriha: Absolutely. and I think definitely, the siloing is something that we've talked about in data for many years. I would say. when I first started working in healthcare, I was surprised by the amount of messy data we had to deal with. but again, like I said, like this is, this is a persisting problem forever.

and I'm, I'm happy to say that I have, you know, been instrumental in solving the problem in many different organizations in building a enterprise data warehouse. Which is sort of the one shining store for all of the data that is needed for analysis. But it turns out that, you know, again, I think the key lesson here is, while that might be true in that. At that point in time, it changes daily. And so really the key lesson to me [00:11:00] is own the chaos. again, and then think about what can we do about it. so let's say that I did end up creating this one enterprise data warehouse. Well, the next, you know. Person who came along said, actually I need this new data source, or I wanted different, different cloud data.

‍

Whereas I don't even wanna work with this pre on-prem that you have because now the world has moved on. healthcare. Or any industry in general will constantly be grappling with new types of data sources, newer or vendor migrations even, which means that now we are switching things around and have to think about how does the past marry with the future. so this is an ongoing problem. so key lesson one to me is just like, own it under like just. Absolutely accepted. and then part two of that is filter for what is most important to the organization at that point in time. So obviously, in any [00:12:00] organization, including in healthcare organization, there are many places where data can be used, but I think the trick is in finding the opportunity that will move the needle the most at that point in time. So this is where like. A good data leader will think about like, what are the organization's business objectives for that point in time? What are they trying to do? What is the biggest goal or the reach at that point in time? And then what is the biggest pain point that is, you know, impacting the ability to reach that goal? and so how can we leverage data to try and solve that particular problem? And. Do we have even access to the right type of data, to solve that problem. the other thing I see, in general data professionals, particularly analysts suffer from is, the need to have perfect data. I, I. know, dealt with this myself and had to like, work, [00:13:00] work with it. I think we have to, again, live in this messy world and understand that we need to be directionally accurate. We don't need to be perfect because we, we'll never get there and we'll be, you know, stuck in this analysis, but analysis if we continue to do that. So really, like if there was a new, data leader, I would tell them, first of all, figure out where do you want to focus and then. sure that you have the data you need to focus in there. you know, make sure you can get some cohesive story, to be able to help the organization. I can give some examples of, you know. How in healthcare organization might think about this. So for example, think about a healthcare organization which is maybe struggling with operational inefficiencies. So maybe they have appointment access issues. And this, is very in healthcare, right? If you try and go to find a appointment with a specialist, like they'll say, oh, we'll see you in six months. Or actually, we're not accepting new patients [00:14:00] right now. terrible. Patient experience. So, how can they get a team help? So the data team can help by thinking about. what does appointment data look like? Are we seeing a lot of news shows or, you know, what are the relevant data sources? First of all, do we have data from scheduling? Do we have data from a call center? Do we have data from the check-in system that the hospital or organization is using? Can we identify any gaps or patterns or bottlenecks in those data sets? And then using that, can we tell a cohesive story and at the end of the day, recommend an intervention? To optimize how appointments are being created, or can we create targeted patient reminders if, for example, if there's a, you know, very specific cohort that's constantly missing their appointments. again, like to summarize, I think data can do a lot, and the biggest and most important pieces of the business is where the, their team should try and focus, and become a proactive force rather [00:15:00] than, you know, always being reactive to needs.

‍

Anthony: So I, I love that idea of owning the chaos. 'cause I think there's this common tendency with, in particular with data leaders, that they imagine that there's some future state they're working to where everything's. Figured out. And, to your point about the data warehouse, it's like, yeah, once I get everything in the data warehouse, then I can stop worrying about this and all the data I need will be there.

‍

and that point never comes. There's never this point where, 'cause it is, even if it were achieved, which probably couldn't be even if it were. One only need wait a beat, and then some new source will show up or some new, modification of the data will be required. So I think this is a, a kind of a, a key point, but I wanted to push you a little bit on, the second thing you said, which is connecting to the business strategy.

‍

And I think one of the challenges I've seen data teams struggle with is too. Understand the business [00:16:00] strategy and, and like I think once people understand the business problem, they clearly have good understanding of the data. Connecting the two is something generally data people are good at. organizationally, how have you worked to make sure you are close to and understand the business requirements and strategy?

‍

And a little to your point, as much as the data's always changing, the strategy's always changing too. And again, that's not a failure, that's a success. Like we want things to change. But, Going Back to the core question, like connecting closely to the business. What's the thinking there?

‍

Spriha: Absolutely. I think this is essential for the data. This is the part where the data leader either shines or not,

‍

Anthony: Right.

Spriha: can they be embedded in all the cross-functional forums that are relevant? In deciding, you know, the organization strategy, are they bringing their, you know, strategy hat to each of these conversations?

‍

Or they're thinking brainstorming hat of like, I can get you data here, or I should get you data here because this seems absolutely relevant [00:17:00] for us to move forwards as a business, and then bringing that back. To the teams and making sure that that work is being prioritized above and beyond any other, you know, growth tasks, solidarity might be, working towards.

‍

Anthony: I think maybe to say it very plainly, being in the room, like be part of the conversation. and maybe that's also. The, the counterpoint is also relevant, which is if you feel like you're not in the room, then perhaps that's an organization that either, either doesn't value and see value in the data, or as a leader, you need to figure out a way to, to escalate and get, get in the room.

‍

Spriha: I think that's right. I also think, you know, good analysis begets better analysis or more analysis. So if you've shown that you can move the needle right with data, then you will be in the room. why would they exclude you?

‍

Anthony: Fair point.

‍

Spriha: it's, it's really about being, maybe, maybe if you are thinking about starting [00:18:00] from scratch, right?

‍

Like, did you pick the right priority to begin with?

Uh, if you did, you showcased your value and then you will be a part of the room.

Anthony: 

that's a great point. I, I think this is true of, career progression as well. I often make this point, don't get the job and then do the job. You do the job and then get the job. [00:19:00] So, you know, if you want the, the promotion, then act as though you have it and all of a sudden it will become obvious that the person who should have the promotion is you.

Spriha: I completely agree.

‍

Anthony: So let's shift gears a little bit. I mentioned in the introduction these two big and slightly confusing terms, and I know you, you have both strong opinions and also thoughtful opinions on both. So let's dig in. this concept of data products. Is, you know, it's one of these things that seems so simple.

‍

it's really just putting two words we like next to each other, data and products. but I also feel like has no clear definition and I feel like there's this sense that, people say it and then they expect the, the listener to understand it to mean something. And in fact, everybody has a slightly different take on what it means.

‍

so. what, what does it mean to you and, and is it useful and, and relevant in your work?

‍

Spriha: Absolutely, and I think this is a very [00:20:00] interesting and relevant question that came up when we were having, the, you know, prep call even for this

‍

podcast. in my world, I use the term data product to imply any kind of analysis or even data. That is used to derive value by a stakeholder. Now this could be the data itself.

‍

It could be line level data that maybe somebody's sending somewhere and we can talk about that. Or it could be insights that are derived from data and it doesn't even, you know, necessarily actually mean rows and rows of data, which you and I as data people would think about. So let's talk about the different types of data that is currently. You know, offered maybe as a product in the market, so. To me, the first type is line level data, right? So this is data that you might buy from a vendor to do an analysis. So for example, pharma companies are out there [00:21:00] buying data to understand how drugs are being used. That are in the market. Maybe they're evaluating competitor products and how they're doing against their products.

‍

So that is really at the line level. then there are the types of data products which are models. So these are not just analysis, but these are predictive, predictive models that are built. On top of some type of data. and in the healthcare Conte context, this could mean a risk score or a readmission risk score that is, sold by several, vendors.

‍

and hospitals often buy those. a third type of product, is a analysis or a data analytics dashboard of some sort. So in some cases you can actually purchase data in the form. Of a dashboard, like a BI tool, or even a curated analysis that helps you find. Your data story or your [00:22:00] answer without spending too much time trying to do the analysis or, you know, investing in an analyst who will do the analysis for you. and it can be very interactive or not depending on what you know, what you're buying and what you're thinking about and how it is packaged. It could even be an application. So there are these data apps out there. all of those to me are data products because. somebody's deriving value from them and you know, they're being offered in the market.

‍

Anthony: I like that. the link to value that I think is a, a defining feature of what makes a data product different than data. It's the fact that someone's connected it to some use case or problem that they're working on. You, you mentioned towards the end of that, this idea of models,which I think brings me to, another important definition that I think everyone struggles with.

‍

it's, impossible these days to talk about data and analytics and not talk about ai. And yet it's [00:23:00] also one of these very ambiguous terms, and I think for someone like you, and possibly for someone like me who's been working in this industry for quite some time, you know, we've been building predictive models, machine learning algorithms years, and yet.

‍

All of a sudden, of course, AI's having a moment. Everybody, as I think conflating AI with generative ai and in some cases in the layman, maybe even conflating with chat GPT. so how do you think about it and how do you, in particular, how do you help the organization think about it?

‍

Spriha: This is such an important problem, especially for, data folks like us who think about ai, like not as a new tool, but something that's been around forever. so again, like let's go back to bras tacks and think about ai, right? What is ai? AI is artificial intelligence in general. It is the science of making machines perform tasks like humans. In healthcare, the concept of leveraging AI has been around forever. by forever, I mean like at least 20 plus [00:24:00] years. So for example, we've been talking about clinical decision support, for a long time. So we use. Drug interaction checkers. we predict risk of certain outcomes. We have these diagnostic support tools which try and help us understand the symptoms that a patient might have. All of this is ai. and so, then the question becomes like, how do we differentiate the different types of ai? so I think, like I said, AI in general is just using. Machines to perform tasks like humans. There is a subset of AI that is called machine learning that also has been around forever.

‍

And essentially machine learning enables software applications to become more accurate at predicting outcomes without being explicitly programmed to do so. Right? So this is the, risk score, for example, that I talked about, because it looks at a number of inputs. It says, oh, I think the likelihood of an [00:25:00] outcome is higher.

‍

This is where the machine learned, looked at a bunch of inputs and had, know, a weightage or a score on each of those and said, okay, these ones matter. These ones don't. and so that's machine learning and that's, again, been been around for a long time. next tier or subset of ai, even within machine learning to me, is deep learning. So this is where machine learning models. Mimic brain structure, or what is called neural networks. So there's a very special type of machine learning which tries to think or use the, brain structure that humans have. Generative ai, it turns out, is actually a certain type of deep learning model that is used to generate. Content. So this content can be text-based, it can be a image, it can be a video, or other things like that. I will also say large learning models are a subset even of [00:26:00] generative ai. and so this is where we sort of like getting into what is chat, GPD, Large learning models are the ones that specialize in processing and generating text-based content specifically.

‍

And I know it's all sort of like touchy butty or all gen AI at this point, but really it's very, very, each, each of these is, you know, a subset of the other, and what we currently call AI is essentially. A large learning model that is designed to understand props that are like human language. and that helps it predict what is the likely sequence of words.

‍

Right? So that's basically what. The model is doing, it's trying to mimic human language. and it does use a deep learning architecture. So again, it's like a subset, to process and generate new content. I think the conflating of all of these terms is absolutely problematic. And the best way that I found to navigate it is to go through [00:27:00] this definition, in every conversation I have, around AI with, you know, teams, or organizations.

‍

Anthony: So I, I totally agree, and I also very much agree with the framing, which is that this, this top level construct of AI is very broad and, and this idea that generative AI and, and heaven forbid chat as a particular mode of interaction with that generative AI is a very strict subset of the broader concept.

‍

And so I think I very much agree with the framing you have. And, I Do agree that there's a challenge with lay people in particular when you have people using, you know, consumer applications like Chat, GPT or Gemini or co-pilots and their, having these experiences and thinking, oh, this is, this is ai, but let me, if I can push you to.

‍

put this in context of the work at Ophelia, which I think is, super fascinating. and, but I think this will help kind of bridge this definitional question with the practical reality of how do [00:28:00] we apply AI in healthcare and help to create better outcomes for patients and. very briefly, and I'll let you expand on this of course, but you know, the, the work of Ophelia is in this very complicated and sensitive area of, of healthcare around opioid use treatment and, in particular for, you know, very vulnerable patients.

‍

you know, this really bumps into this idea of how to scale the delivery of opioid treatment care. and the use of ai, and again, you can expand here of course, but my sense of AVI is it's trying to use AI techniques. And again, I use the broad AI construct here, to help achieve that, I think very important outcome.

‍

But you tell me like, how do you see it in the context of the work at Ophelia?

‍

Spriha: Yeah, it's a great question. and again, like we can, we can talk about all types of AI here, but, I will restrict to the two types, two higher and lower types [00:29:00] just to, keep this conversation a little bit. shorter. so at the highest level, I talked about machine learning, right? And so that has had applications in healthcare for a long time.

‍

So I talked about no show risk, right? So we've, you thought about that. Like, we see some patients are constantly no showing. Can we, predict that and then can we try and, you know, intervene in there? That's one. Potential example. another one is we see that a number of patients are maybe, you know, at higher risk for a certain type of outcome.

‍

Maybe they're leaving, us too quickly or maybe they're higher risk for certain. outcomes that we don't want them to have, and so can we find them ahead of time. in one of my previous roles, my job was to try and come up with a risk model to predict future high utilization. So that's the ml, you know, like traditional AI that again, has been around for a long time and that we [00:30:00] use, And then going down to now gen ai, which is obviously, you know, the, the topic of the day, so to speak. there is a lot of use case, honestly for Gen AI in healthcare as well. in any industry really, the use case. For any technology, not just gen AI should be, how can I make my job easier and more efficient?

‍

Right? So, as an employee, I want to do the things that excite me. I don't want to do the things that, you know, are like rote or routine, or something that. a tool can do or a software can do, or a machine can do. And so this is where AI comes in. Back in the day, it was where software came in. so how can AI help in healthcare, particularly, if you ask any clinician the part of their job that they want to replace, but Overwhelmingly is the documentation. So back in the day, clinicians were taking a hundred and notes. They were, you know, prescribing [00:31:00] meds and signing orders, all, all on paper. then came, you know, the ehr, electronic medical records, and now there's a computer screen and between a patient and a clinician. And there is a lot of documentation around clinician burnout with an amount of documentation that is needed, or Patient dissatisfaction with just, not being able to make eye contact with their provider when a lot of what is happening in the, in the medical encounter is human, and requires that eye contact. And so this is where there's . Absolutely a use case for ai. and we talked in the beginning you mentioned about healthcare data being unstructured. a lot of these notes are, you know, text paced and this is where gen AI or large language models shine. They understand human language, it turns out and can summarize a lot of what's going on. There could be, and there's so many tools out there. some we've even played with ADO internally. but you know, you, [00:32:00] there could be a. Gen AI listening, converting voice to text while the conversation is happening so the clinician doesn't have to be typing. They could be a, a bot summarizing what happened in that conversation or in a series of conversations and events. It could be that. even between visits the clinician or had some questions for the patient or the post patient had questions for the clinician and so they texted in or they had a phone call and there are transcripts that the, LLM can read and summarize and make it easier for the clinician to understand what happened even while they were not in the room. so again, this is, this is where I think, you know, we can use gen ai, a lot more. Than, than we ever have before to make the lives of clinicians better. The other part I will say, is billing in healthcare. which we talked again, talked in the beginning, about being extremely hard. And that's the other very clear place for us to leverage [00:33:00] LLMs, to get, you know, better, RCM software. I think I read recently, that there are eight, unicorns in healthcare tech, particularly in ai. and they're all within the RCM or the text summarization domain, so that just tells you where the value is.

‍

Anthony: Interesting. Do you think there's something specifically relevant about, opioid care that that sort of lends itself to predictive models and maybe even to, generative, Techniques, is it, is it your view that there's something specific about, the way that, patients interact in the opioid, care space that makes it specifically interesting in, is there something different about that?

‍

Spriha: I think what I can answer from the, model that we have is. see patients a lot more frequently than in, you know, like a traditional practice. So even if the patient has an [00:34:00] acute, you know, crisis of some sort, they might have a series of visits with a provider and then they'll drop because, you know. It changes. But in OOOD we actually need to see the patient for a long time, to get them, back to their lives and, consistently. So that is where not only do we care about what happened in that visit, but really we, we wanna know what happened to the patient on the whole. and that's where I think, like I said. LMS can have a role because we can really do a much better job of understanding what's going on, in their lives. we haven't done this at all yet, but we talked about the iot and you know how there are all these, Smartwatches and phones talking, looking at what is happening daily, in your life.

‍

And I think that is another opportunity for us to, you know, use LLMs and bring all of that data together to really tell a cohesive story of what's going on in the patient's life in.

‍

Anthony: So, would it be fair to say that [00:35:00] one of the unique challenges in opioid care is that it's simply a lot of data, a lot of very varied, very varied data. and as such, the challenge associated with any data task, whether it's summarizing, predicting, is more challenging.

‍

Spriha: I think that's fair. I think that's true for healthcare in general,

right? Like you can get as much data as you want, and get better at understanding what's going on in the patient's life. I think what's different about OUD is that it's so persistent and, you know, you need to see your provider every month or every, whatever the, cadence is based on your, requirements as a 

‍

Anthony: Well, Thank so much for the time today. This has been a great conversation. I think we've, covered a lot of territory, and, really appreciate you making the time for us.

‍

Spriha: Thank you. It was lovely chatting with you.