Addressing Social Determinants Before They Become Chronic Issues with Marty Elisco, co-founder and CEO of Augintel

[00:00:00] .

[00:00:02] Hey, everybody. Welcome back to the Outcomes Rocket. So glad you tuned in to another episode because today

[00:00:09] I have the privilege of hosting Marty Alisco. He is the co-founder and CEO of Og Intel.

[00:00:16] It's a SaaS company that improves the ability of health and human services professionals to deliver

[00:00:21] care. Og Intel builds natural language processing software used by caseworkers,

[00:00:27] therapists, and clinicians to identify key information buried in text in case management

[00:00:34] and medical record systems. We all know it's a huge problem. This gives them a deeper understanding

[00:00:39] of the at-risk individuals and families they care for. And I'm excited to have him here to chat

[00:00:44] about it, what they're doing, how they're doing it, and how it's making a difference. Marty,

[00:00:48] thanks for joining us. Thanks for having me today. Of course. Now look, always curious

[00:00:53] on the founder's stories. What is it that got you into healthcare?

[00:00:57] Well, I come into healthcare from, I think a pretty different avenue from the social services and

[00:01:03] technology arenas. I started my career in telecommunications and about 15 years ago

[00:01:10] had an interest in bringing technology to the child welfare and social services space. And

[00:01:15] ever since then have built different businesses in this space. But about

[00:01:21] five years ago, maybe six years ago, we realized that there's a huge need for

[00:01:28] identifying language buried in text, especially in the child welfare and social services space.

[00:01:33] So we built a company that did that to help child welfare and social services organizations

[00:01:38] understand that people they care better. But it turns out the same information that

[00:01:44] we're identifying in social services has a direct impact in the more traditional healthcare

[00:01:48] space as well, especially in the area of social determinants of health. In the child

[00:01:54] welfare and social services space, the social determinants are a key cause of the health of

[00:02:00] individuals, right? And it turns out in the entire population that is as well. So I came

[00:02:06] at healthcare from really understanding the world of social services and bringing some of the

[00:02:11] sort of solutions to problems in that space into healthcare. The healthcare world is known

[00:02:17] about this, but really bringing that lens of social care and well-being to the physical

[00:02:22] health arena is really how I arrived at healthcare. Super cool. Yeah. And they're inextricably

[00:02:28] connected as you mentioned with so many of these social determinants affecting people's lives

[00:02:35] and their health. The idea that zip codes are more determinant than the DNA code.

[00:02:40] So tell us more about Aug Intel. How are you guys adding value to the ecosystem

[00:02:44] with the technology, the services you provide? So really importantly, like I said before,

[00:02:50] everybody recognizes social determinants of health are a huge cause of a person's health.

[00:02:56] The problem is the tools today don't really have the specificity needed to identify what

[00:03:07] social determinants affect specific individuals. So before you mentioned zip codes and there

[00:03:13] are a lot of platforms out there that can use publicly available information to

[00:03:18] infer what problems a person has based on where they live or certain characteristics and demographics.

[00:03:26] But what we do is something much more specific. What we do is look into the patient's data,

[00:03:33] especially the narrative data for language that describes those social determinants,

[00:03:39] because we believe that in that language are the sort of truths about what affect that individual.

[00:03:46] The social determinants we found are predominantly described in that narrative,

[00:03:53] in that qualitative unstructured data, not in the quantitative data and not in publicly available

[00:03:58] information. So what we do is use natural language processing to mine narrative data for

[00:04:08] content that describes social determinants and then present that information to clinicians in a

[00:04:14] wide variety of ways so that they can act on it. And I'd like to literally list the social

[00:04:21] determinants that we look for, which is really the sort of global world because a lot of times

[00:04:27] it'll be EMR platforms that pick out three or four of them. But you've got to look at all

[00:04:32] them to understand what's affecting the patient. And so we have food insecurity, housing instability,

[00:04:38] homelessness, inadequate housing, not just homelessness, transportation insecurity,

[00:04:44] not being able to get to a doctor's appointment, financial insecurity, material hardship,

[00:04:49] right? Educational attainment, veteran status, stress, social connections, intimate partner

[00:04:56] violence, mohistan elder abuse, health literacy, whether or not you're covered by health insurance,

[00:05:02] medical cost burdens and even digital literacy, right? With EMRs today, the EMR really expects

[00:05:08] the patient to be able to navigate their own way through a healthcare system. And if you have a low

[00:05:12] degree of digital literacy, that's very difficult. So what we do is mine notes for every single

[00:05:19] one of those so that the clinician can one, there's so many different use cases for this

[00:05:24] data, which I guess we can get into. But I'll leave it at that, which is that's what we're doing,

[00:05:28] right? And manifesting this information that really has been a blind spot to clinicians,

[00:05:34] therapists, care coordinators forever. Matt, thank you for giving us that very clear value

[00:05:40] prop you guys give now you're working with, I imagine providers and provider organizations,

[00:05:46] right? Mainly behavioral health organizations starting to work with value based care

[00:05:51] providers. And of course, like I said, we work with many social services and child welfare

[00:05:55] organizations at the government level, state, local, as well as nonprofits.

[00:06:02] Cool. So a lot of FQHCs in there as well that you're helping?

[00:06:07] Yes, we can talk a bit more about FQHCs. We do believe that we will benefit those types of

[00:06:14] organizations, but we really haven't been able to crack that code yet.

[00:06:17] Got it. Got it. All good, man. No, thank you for that. Let's talk use cases

[00:06:21] in sort of ways that you guys are adding value. Yeah, so I'll focus primarily on the behavioral

[00:06:27] health and value based care space because that's where it's tangible. So if we're looking at

[00:06:31] value based care, everybody knows that the way to optimize outcomes is to address

[00:06:38] the social determinants and social well being so that they don't manifest into chronic

[00:06:45] health conditions, right? So what we do is give the value based care organization,

[00:06:50] or really any provider as well, the ability to identify the specific social determinants that

[00:06:58] are affecting that individual so that they can refer that patient to the right community

[00:07:03] based service to address that problem. So essentially we are the tool that helps to

[00:07:10] prescribe social care to that individual so that the value based care organization can

[00:07:16] really optimize their health at a lower cost, right? Community based services are of course

[00:07:20] less costly than and frankly more effective earlier on in a patient's care than getting to

[00:07:28] the ED or dealing with chronic issues. So that's the first piece, just identifying what to do.

[00:07:34] The second piece is very much around risk stratification, right? Value based care

[00:07:38] organizations need to prioritize which patients they need to focus on, right? Problem is the tools

[00:07:44] today to help risk stratifier really only based on that quantitative data and you get,

[00:07:52] I've heard this over and over again, we have an organization where two-thirds, not sorry,

[00:07:57] I'm exaggerating, but like 20% of our population is as a highest risk score,

[00:08:01] but we just don't have the resources to support that. And the reason is they don't have

[00:08:06] visibility to the granular conditions around social determinants that would help them to

[00:08:12] better risk stratify and identify the sickest patients or the patients in most need of help.

[00:08:17] So what we do is give them the information to help them create a much more robust risk model

[00:08:25] so that they can stratify patients into the right tiers of care, into the right categories

[00:08:30] of care that help them optimize what's delivered to that patient. Interestingly enough,

[00:08:38] from a revenue perspective, right? We're starting to see with when I listed all the social determinants

[00:08:42] before, there's Z codes, right? There are actual ICD and SNOMED codes that are focused on

[00:08:48] social determinants. And if an organization can code for those, right, they are eligible

[00:08:55] for higher reimbursement rates. And because, right, they're showing the true health of that

[00:09:01] individual and of course the true health of the population. So enabling organizations to

[00:09:07] better communicate the health of an individual by incorporating social and mental issues

[00:09:14] into the patient record, it helps them from a financial perspective as well.

[00:09:20] Totally. Yeah, risk stratification, revenue possibilities by being able to really get a hold

[00:09:27] of what the problems are using the right codes for them. You guys are helping in a lot of ways.

[00:09:32] And so as you've approached this problem set in healthcare,

[00:09:38] what's been one of the biggest challenges you've dealt with?

[00:09:40] Yeah, a few things. We're an AI company, right? So I'll start with the technology

[00:09:46] then I'll start with part two around the practice side of it. From a technology perspective,

[00:09:51] the first problem we had was is the language model accurate enough, right?

[00:09:56] And really what it takes is using the organization's own data to train the model

[00:10:01] to make it smart enough to understand how that organization does business. So

[00:10:08] we mitigate at risk by when we onboard the organization, we train the model to their

[00:10:15] data and that is shown to give them the regionally specific, the community specific

[00:10:20] data that they need to have a model detect what they're looking for. A social determinant in one

[00:10:25] community could be described completely differently than a social determinant in another community.

[00:10:29] So training on their data is what addressed that problem. And again, we're an AI solution.

[00:10:36] So there's just generalized mistrust of AI, especially generative coming out, right?

[00:10:43] We've been doing this for a long time, right? What we do is really, I would call it the first

[00:10:47] 80% of generative which is identifying what's important in the data. The last 20% is reconstructing

[00:10:53] that data into sort of a summarized view or what we call abstract of summarization of that.

[00:11:00] And so the first 80% we've been working on forever but with generative came all these

[00:11:05] opinions on the risk of AI in this space. So we've had to just be very clear that our AI

[00:11:13] doesn't make decisions. Our AI presents the clinician with the information they need

[00:11:18] to make that decision to avert where people go with AI, which is get to these sort of

[00:11:24] nightmare scenarios around it automating when the decision making process of clinicians.

[00:11:29] So as long as we're very clear that we're just pointing you to the data you need to

[00:11:33] be aware of to make decisions, it really sort of addresses that issue.

[00:11:39] And then from an outcomes perspective, when we're talking with value-based care

[00:11:43] organizations and other physicians, now I'm getting to the outcomes problem.

[00:11:49] Many times clinicians are reluctant to identify social issues and sort of state them in the

[00:11:56] primal list because that organization has nothing they can do about it, right? There's not a

[00:12:02] robust referral network to community-based organizations that can address that issue

[00:12:07] or the health network care coordinators don't have the bandwidth to address them.

[00:12:12] So this is changing as people become more aware, but a lot of times people say,

[00:12:19] I know this is a problem, but I don't have anything I can do about it, right?

[00:12:24] I don't have the resources in my network to be able to address these problems.

[00:12:28] So it becomes a boondoggle, right? We have to get over that. Not just us as a company,

[00:12:32] but the health care community needs to realize that you got to figure out a way to address

[00:12:38] social determinants with community services to maximize outcomes in your community.

[00:12:44] And this can be healthy communities. It can be at risk communities. It doesn't really matter.

[00:12:48] This affects, I described the list before, these issues affect every single community

[00:12:52] in the United States. How do we bridge that gap? When the organization is not incentivized,

[00:13:00] just directly, the incentive is keeping people healthy, right? So the incentive is there,

[00:13:04] but there's no reimbursement and everybody's burned out having to put stuff into the EMR

[00:13:12] waiting room full of patients. How do you incentivize that physician to get that care

[00:13:20] at the community level? And then the second question that I had for you, Marty, was really

[00:13:24] around visibility. So what you're telling me is you guys are giving this information to the

[00:13:30] point of care of the physician. What opportunities does the organization have at a population

[00:13:35] health level to take a look at this? Yeah, okay. So the first question is,

[00:13:39] what does the health community need to do to address this? First, I think this is built

[00:13:43] in the value-based care, right? So I'm a strong supporter of us as a country moving more and

[00:13:51] more towards that model because I think the incentives are absolutely there. And

[00:13:58] these organizations are getting better and better at addressing them. But I still

[00:14:03] thinking on regular fee for service models, there is incentives to improve health, right?

[00:14:10] There may not be financial incentives, but I think that there are the physicians who want to

[00:14:18] do the right thing, right? And that's a harder hill to climb. So again, I think the solution is

[00:14:25] moving to a value-based care model. And then remind me what the second question was.

[00:14:30] Yeah, the second one is the information output available at the point of care versus

[00:14:35] at the population health. Yeah, when you look at an organizational level,

[00:14:39] right? And I didn't really get into this, but our product not only looks at individual patients,

[00:14:44] but looks at trends across an organization or across a community. So first,

[00:14:52] resource allocation, right? If you can look across your population and see what the true

[00:14:58] social determinants are that are affecting individuals or particular communities or

[00:15:03] particular areas, you can know how to allocate resources at the health level or

[00:15:07] to even embed community services within the community. This gives you that instantaneous

[00:15:14] market research to understand at a specific sort of granular level where to put what

[00:15:21] to maximize health outcomes and also from a quality perspective, right? You want to make sure

[00:15:27] that your care coordination teams and your physicians are doing something about it, right?

[00:15:35] There's identifying the issue, but then there's what happens. So we have the ability to look at

[00:15:41] sort of like if this then that, like if this patient has this scenario, then what was done

[00:15:45] about it? To look at that from a social care perspective so that you can see

[00:15:50] how your staff is performing. Do you need to be training them differently? Do you need to be

[00:15:54] better networking with the community services that are already out there, right?

[00:15:58] Do you need to be creating new workflows because they're bottlenecks, etc., etc.

[00:16:03] So that's the sort of thing where you start to implement this at scale by looking at the workflow

[00:16:09] of this within your organization. I appreciate that. Yeah, I just wanted to make sure we captured

[00:16:14] that value offer because it's a big one. Cool. Marnie, super interesting. I love the work

[00:16:20] that you guys are up to for everybody listening. What call the action would you give them?

[00:16:26] What closing thought would you leave them with as they consider the things that you

[00:16:30] guys are up to? Well, closing thought from a practice perspective, I'll say it again, right?

[00:16:36] Figure out how to address social determinants before they become chronic issues with your

[00:16:44] population, right? The data's there, right? It's all there, right? It's finding the right

[00:16:49] technologies and tools to make you aware of it so that you can do something about it.

[00:16:55] And with the growth of AI, with the ability of AI and NLP to be tailored to specific use cases,

[00:17:06] you now have the quality and accuracy to do it properly. In second, obviously, selfishly

[00:17:13] speaking, right? If you want to talk with us about this, you visit our website at Augentel.us.

[00:17:17] We make it pretty easy. You can book a meeting with us on our website to talk through it,

[00:17:22] and we're happy to go deep, go shallow, go as wide as you want and teach you about how NLP in

[00:17:27] this space works, explore how it may help your organization or just have a nice quick conversation.

[00:17:34] So anyway, works for us. Love it, Marty. Really appreciate that, folks. Take advantage

[00:17:40] of Marty's offer to connect here. Certainly he and his team run a really great business with

[00:17:47] software that could really help you identify ways to operationalize care treatments for those

[00:17:53] social determinants of health. Take advantage of that. We'll leave everything in the show notes

[00:17:56] for you to get in touch with Marty, check out Augentel and all the amazing work that they're up to.

[00:18:01] Marty, appreciate your time here today. Thank you.