The State of AI in Healthcare and Pharma

Introduction to Spencer Honeyman

John Farkas:

Well, greetings everyone, and welcome to Healthcare Market Matrix. And today we are going to be diving into the wild world of artificial intelligence. And I know everybody is diving into the wild world of artificial intelligence right now, so why not us? But here’s what I want you all to know. I was just at the VIVE conference, I was at HIMSS, I know that every panel discussion or every other panel discussion or maybe 60 something, 70 something percent of the panel discussions at both those conferences had the word AI or the acronym AI in the title, and every booth, practically, had it represented in some way, shape, or form.

So it’s already achieving some level of ubiquity, but does that mean it’s really achieving some level of ubiquity? What’s really being done? How is it really being used? And I think the answer to that is, it’s certainly not quite yet. There’s a whole lot of gaps to close, a whole lot of opportunities to realize. And in a lot of ways we are really just at the dawn. It’s a very big topic and the topic of a lot of conversation right now.

And we are here today with Spencer Honeyman, who’s the Chief Revenue Officer of Vi Technologies, and they’re using artificial intelligence to enhance the outcomes and financial returns of healthcare organizations. And what they’re doing is leveraging AI, a suite of AI powered really, they productize some strong data science to help really optimize the customer lifecycle engagement in healthcare. And I’m probably, you’re going to do a much better job of unpacking that for us, Spencer. But these guys have just put a report together that’s basically outlining the state of affairs related to AI right now in healthcare and pharma. And so we’re going to dive into that today and look at some of what that is. But before we jump in, first of all, Spencer, welcome, and I would love for you to tell us a little bit about your backdrop, what brought you to this moment, and what’s got you here.

Spencer Honeyman:

For sure, John, and thank you for having me. Excited to chat with you and be able to hopefully interact with this audience. My background is I am a startup guy. I don’t know if it was intentional, but this is my third startup at the intersection of health and technology. I’d say both personally and professionally passionate about how can we use innovation to help improve health outcomes, ultimately, and help large businesses by doing that capture some of that value. And so prior to V, I was leading commercial growth for two startups, one in the AR/VR space, which was really interesting and is another evolving landscape.

John Farkas:

Absolutely.

Spencer Honeyman:

And prior to that, was in the hardware software was a company that used what’s called near infrared spectroscopy, or light through the skin to measure biometrics and different things going on in a person’s body. So I won’t fully nerd out on that front since we’re going to nerd out on the AI side, but I’ve spent the last five years with Vi and the company’s been around for eight years. And before ChatGPT was a dinner table conversation, we’ve been looking at how to help large health organizations better utilize their data and create configurable capabilities that technologists and marketers at those companies can use to improve, basically, the key KPIs across the member life cycle and ultimately to help improve health outcomes.

A Deep Dive into Vi’s Recent Report

John Farkas:

Awesome. So help me into, so you guys put a report together here recently.

Spencer Honeyman:

Yes.

John Farkas:

If you were to summarize what that report, what the mission of that report is, how would you characterize it?

Spencer Honeyman:

Yeah, so I think what’s a little bit unique about Vi is we cover healthcare, clinical healthcare, I think some of the largest national and regional health plans and disease and care management with high risk, high cost populations to pharma, whether it’s rare disease or it’s a broader vaccine, to wellness, where we’ll work with a direct to consumer membership or subscription service for fitness, nutrition, mental health. And so we’re looking, we say health, we look across that whole spectrum.

And I think what we try to do internally, and we shared with the world is, where do we think AI is at today? And I think we start by I think picking apart what does AI really mean? And I think there’s different pieces to AI and what’s here today, how can you use it, what is not here, yet? What’s the pace it’s moving at? And then what are the trends we’re seeing in the space? I think what’s interesting, to your point of the AI being slapped on everyone’s booth at the big conferences is, if you look at AI projects or data science projects, still, over 80% of them fail their business objectives. And so I think there’s rapid pace of change. We’ve obviously had different experiences than that 80%, but I think what it means and how it’s used, that’s the goal of this report, to share some wins, some scars, and what we see coming.

Exploring AI Opportunities in Healthcare

John Farkas:

So in the context of what you guys report, as much as it’s dominating the headlines, like we just talked about, it’s everybody, it’s everywhere. The implementation has not been that widespread and what you showed us, 35% of companies have successfully incorporated AI commercially, and like 20% of healthcare organizations have adopted some form of AI. So what do you see continuing to shape the landscape of healthcare and other sectors in the coming years? And what are some of those AI, those opportunities look like for AI right now? How are you looking at that horizon?

Spencer Honeyman:

For sure. So I think one thing we see is, AI is not a thing that you’re just going to be able to plug in across every industry. I think that’s one thing we would say. So I think being vertically focused on using technology, whether it’s AI or otherwise, especially in health, where behavior change is such a hard problem, is critical. And that’s why we choose to focus on that space and we think others will need to, to deliver effective solutions. And I would say we pick apart AI into three different buckets. And this is in the report more eloquently than I’m going to put it verbally, but I’ll do my best.

So I think the first piece is predictive analytics, which is not new, which is how do you use de-identified historical data on demographics, psychographics, different engagement activities, different health populations to predict, in our case it’s different engagement, acquisition engagement or retention patterns in the future, whether it’s a diabetes management program, it’s a health clinic or it’s a pharma vaccine or clinical trial. And so we believe predictive analytics is here and it’s being leveraged and it gets locked into that. I think where we would then move to machine learning would be, well, if predictive analytics helps you use historical data to predict future behavior, well if I know someone is going to disengage from my program, what do I do about it?

And that’s where you can test interventions, whether it’s a call, an email, a text, it’s a reward, it’s an offer, it’s showing up at someone’s doorstep. And what machine learning’s really good at is basically, A-B testing on steroids. How do you test what intervention strategy on what channel at what time works best for different groups of people, and even people on a one-to-one level. And that learning is a lot more sophisticated, faster, and allows for personalization. And so we would say that is here and we’ve seen it in the market very successfully implemented through our customers and others not using our products and services.

And lastly on AI, I think that in our lexicon, that would mean more automation and more generative AI where you put an input and you expect them to generate the email template, the advertisement that would go on Google or Meta. And so what we do at Vi is we’re not in the generative AI space now. What we do is we provide frameworks to marketers and product and technologists teams at our customers to use predictive analytics, some data that we license on consumer behavior, and machine learning to optimize their content, their email script, what the nurses should say on the phone, what the automated tech should say. And I think what we believe generative AI is here and coming, but that fully automated robot, like Terminator, experience we think is a little bit far out. So we’re excited about it and it’s moving fast.

John Farkas:

Let’s hope the Terminator experience is more than just far out.

Spencer Honeyman:

Yes.

How AI Has Alleviated Administrative Tasks

John Farkas:

Let’s hope it’s not going to happen that way. Yeah, so we’ve got three decent sized buckets here. It’s the use of how do we get over some of the administrative things, which everybody agrees uniformly. Some of those, the busy work administrative hurdles are some of the low hanging fruit in this realm, and then we have the member experience space, and then we have the predictive analytics space.

If knowing that, let’s dive into the administrative space to start with here. And looking at some of the administrative tasks like scheduling, and claims processing, data entries, those alleviating some of that workload, and for healthcare professionals is some obvious opportunities, so they can focus more on engaging their patients and actually providing the care. So can you provide some examples where AI driven solutions have been successful in alleviating some of those things for healthcare professionals and really doing the work of improving efficiency for some of those workflows?

Spencer Honeyman:

For sure. I think for the vast, vast majority of healthcare professionals, there’s never enough time in the day. And so I think operational efficiency is obviously critical. And so I think a couple of examples, I think for both I’m going to share here, it starts with having data that is in an infrastructure that can access both historical and ongoing data to be able to help these things be automated in a way that works. And so it’s cliche and it’s been a garbage in, garbage out since the days of big data, etc. But I think it does still apply and you need to have a way for the AI or the technology to learn.

I think two really interesting examples. So we have a customer, they’re a very large health system, and they had a lifestyle coaching service for those with pre-diabetes or early stage diabetes. And what they found was the health outcomes were most linked to the compatibility of the patient with their provider, and so, because this was an ongoing coaching process. And so what we found was really interesting was A, how do you match, how do you look at the good matches and “the bad matches” or ones that weren’t a great fit. So you’re able to, in an automated way, figure out, as that patient comes into the system, who would be the right health professional for them to work with. And then what was really interesting was we started to see patterns and we were able to automate some things like scheduling the next appointment at a time that works for that health professional, in general, and matching it to what works with the person receiving the care, the patient.

And what we found was not only better results, but also we were able to increase bandwidth and productivity for that healthcare professional, the number of patients they could serve in this diabetes lifestyle management program, we were able to increase it over 35%, which is obviously transformational, in terms of the impact that organization could have on the population.

Employing AI to Increase Patient Engagement

John Farkas:

Yeah, no doubt. And we’re seeing lots of different manifestations similar to that across the ecosystem with a variety of different solutions. When you are looking at, if we were looking at things related to member engagement and enrollment, engaging members is obviously one of the big challenges that is existing right now. It’s so hard that Walmart can’t even do it, apparently. But as we are looking at the advancement of AI and how what kind of innovation, innovative solutions can come to the table in that regard, are there specific populations or healthcare scenarios where those kinds of interventions are really showing some promise or look really promising on the horizon as far as what we anticipate being able to happen?

Spencer Honeyman:

Yeah, I think what has remained true and what will continue to remain true is there’s this unfortunate correlation between certain health populations that have high chronic risk factors, usually more than one of them, and how hard it is to engage those in the care they need and keep them in that care to ultimately improve health outcomes. But I think what we have seen is that you’re able to create the uplift, in terms of the engagement level of those people if you apply different technology and AI frameworks.

And so to make that a little bit more practical and give you some examples, so we work have health plan customers and they will have certain partners or internal programs that go across, let’s say chronic disease management, so MSK diabetes, kidney disease management, behavioral health and more, and I think we’ve seen a couple things be really effective. So one is I think in a HIPAA-compliant way, there are ways to leverage third party data sets that are available. We do some of this at V, as well, to enrich the data health organization may have, but not related to claims or Rx data. So for that profile of patient that you’re looking to engage, where do they go? What retail stores? What restaurants? Do they go to a health club? What do they search online? What different behaviors do they have when it comes to what they purchase, things of that nature. And you’re able to create more of a 360 view versus just the traditional healthcare only data view. And I’ll give some examples shortly of how that can be applied.

And then I think what we believe is the force multiplier, or what we would call optimization frameworks. So I think we’ve seen from marketers, specifically around engagement, I don’t think going to tell you, hey, here’s the perfect email script you should send to this person, and magic happens. But I think there’s now frameworks where you can leverage the data at your disposal, different third party outside data that is not healthcare related to rapidly test and iterate. So I’ll force you and the audience to nerd out with me very shortly, John.

But I think if you think of AB testing historically, let’s say we have two email templates and we’re trying to get this individual to enroll or engage in a certain program or to sign up for a PCP, whatever the use case would be, and we test A verse B and either A or B wins. I think the difference now with machine learning and AI is you can have A, B, C, D, E, F, G, and in the past where B may have lost to A, B may “lose”, but it might be really good for 3% of the population that’s Medicare in the Northeast that has X, Y, and Z related to them. And we don’t need to throw it out, but we can serve up the right content to the right individual on the right channel. And those subtle tweaks of what works for different people eventually, either in much smaller segments versus a one size fits all or even at an individualized level has huge impacts on whether you’re able to even contact or communicate with your target audience or whether you’re actually able to engage them.

John Farkas:

Yeah, very cool. What are you seeing as far as how eager or open some of the health systems are to making that happen? It’s one thing to talk about, it’s one thing to say we could do this. It’s another thing to actually get the buy-in and permissions and all the things necessary to actually see it through.

Spencer Honeyman:

I think it can vary and it’s certainly, I think, a challenge that the healthcare industry has versus maybe traditional consumer retail of obviously we’re handling very sensitive data. So I think there’s some approaches that have worked better than others. So for us, we’re not in, we don’t predict medical risk, we don’t make care recommendations, what we tell our customers is we can help you find more of your target population, get them in doing more of what you want them to do for longer. And if your care and what you do for them leads to improved health outcomes, that’s our mission and we want to help you do that better. And so because of that, we don’t, in most cases, we don’t need to take in PII, we don’t need to know it’s Bob Smith, I don’t need to know what their A1C was. Maybe we know if they took a glucose reading, and it’s just member 1, 2, 3, 4. And so I think there’s ways to de-identify the data and use more engagement-centric data, specifically for marketing. I think that’s one thing.

I also think second, where we’ve seen a huge change is around value-based care. And so it’s something that’s obviously not new and I’m personally passionate about, but I think the shift in incentives is pretty massive to where organizations are now able to leverage paid media channels. And I’ll give you an example. So let’s say in the past, if I was providing a certain service, let’s say primary care, and my employer, I sell to large ASOs to self-insured employers, and they were covering the cost, in the past where it may have been 2 cents per employee per month, whether they use the service or not, now it’s $250 per engaged member per month.

And while that creates the right incentives, the other thing it does is, hey, now if I can leverage de-identified data to serve up an advertisement on where that individual is today on Google, or Instagram, or Facebook, or Hulu, or wherever it is, I can spend $100, or 200, or $500 to acquire that patient, in the marketing sense, on paid media in addition to traditional non-paid communication channels. And so what we’re seeing customers do is start leveraging both communications and media to meet these audiences where they’re at because the financial economics of some of these value-based care models align.

An AI Success Story

John Farkas:

So Spencer, let’s take the value-based care scenario. What’s a good example of something you’ve actually seen in the market where that has come into play? How has that been realized?

Spencer Honeyman:

Yeah, we saw, we have a pretty transformative case study and this always, this is not a one size fits all, I think that’s for sure with all AI, but we have a-

John Farkas:

That’s what I was going to say, that’s the state of affairs, right?

Spencer Honeyman:

That is for sure.

John Farkas:

Or at least it should be.

Spencer Honeyman:

I think to get into that success using any of these buckets of AI, I would fully agree. I think one example that we thought really interesting was pretty transformative for one of our customers is their regional behavioral health clinic, and they focus on children with rare and pretty severe forms of autism. And they offer this as a covered benefit to large employers and health plans for, usually for their children. So it’s like think ages 2 to 12, or who are the children that have this autism?

And what they really struggled with in the past was they were using communications, and it was really hard to engage the parents or caregivers of these children, explain to them what this is intensive program that they offer. It was multiple times a week for an extended period of time, could provide to them that’s different, maybe, than the care they’re receiving from their primary care or they saw in general, and it was hard to reach them. And also, a lot of times they don’t have permission from that employer or health plan to send the email they want to, when they want to, how they want to. And so using the value-based care example with this shift to only getting paid based on engagement and health outcomes, they were able to start using media.

And so what we did was, with them, their marketing team was able to look at a bunch of analytics of, okay, who are the parents of these children and what do they do? So what are their demographics, age, gender, location, what does their mobility look like? Do they shop at different stores? Do they purchase certain things for their children? That would be very non-obvious, and I obviously can’t go into detail here, but nothing to do with what you would consider a medical device. We’re talking consumer product purchases, that if you have a child with rare form of autism that this would be applicable for, would make you a lot more likely to be in this target audience. What do people search? And again, it can’t be health related, but it could be X, y or Z, Again, without going into too much detail. And do you follow key digital opinion leaders or join advocacy groups?

And so what we helped them do was basically create these segments and use data sets that we have access to, to be, call it 10 to 20 times more precise with who are the mobile IDs and IP addresses without ever identifying these individuals that look like the people that are the caregivers of the parents of your target population. And then they were able to deploy media to those individuals across social search, connected TV, they did some direct mail and they were able to reduce their costs for acquisition over 35%, and their conversion rate went up over 50%. So it was very quality set of leads, which was transformative to them of how do they get the right people in the door, and ultimately provide care in the way that they do, but more economically.

The Awareness vs. The Adoption of AI

John Farkas:

So Spencer, I know that there’s awareness of AI and then there’s adoption of AI, which are two different things. And I know that you, being in the position you’re in within your organization, you are in the conversations where people are confronting some of the hesitations, the concerns, the objections, the fears, whatever those might be. Talk to us a little bit about what you’re seeing in those conversations. What are some of the hurdles that you’re facing that are, I know, if I know one thing, if you’ve sold in the one health system, you’ve sold in the one health system, so there’s any number and they’re varied. But what are some of the common themes that you hear as far as objections or hard, things that make it difficult for them to consider the kind of innovation that we’re talking about here?

Spencer Honeyman:

Yeah, so I think for the most part we’re talking to leadership within these organizations and marketing, product, technology, and data. So it can be a broad group. I think the biggest thing is, is this actually working? And if so, how much and how do I attribute it to this versus all the various factors that can go on in the world. And so what we’ve seen work well on our end is, let’s say we’re speaking to a marketer, and we can go back to the example I gave on, let’s say I’m using paid media acquisition or optimizing communications and email marketing to the right person at the right time through their existing CRM and tools, we would say, “Hey, what would it look like if we could increase your enrollment and engagement 20% or your retention in this program 20%? What would that mean to you as a business, as an organization?”

John Farkas:

Classic challenger sale tactic, right?

Spencer Honeyman:

There you go. And you say so. And if they don’t have an answer, then I think they’ll be like, “Okay, that’s interesting. Let me come back to you and think about it.” And what our model is, and this is after years of pain, and there’s plenty of models that can work here is what we do is we sign long-term agreements with these customers. Usually, let’s say it’s a CMO, that’s our partner that’s actually signing this agreement and we say, let’s create a back of the napkin of what this value could look like.

And everything we do has a control group. So if the world changes, your marketing content changes, you expand, you retract, things go on that are not relevant to what we’re touching, we’ve got a holdout group that it will be a consistent baseline. Not just during a pilot, but always. A, because it’s how the machine learning learns. We’ll still predict on that control group or the holdout group, if you will, but you’re able to attribute the value and the learnings that you’re seeing, whether it’s reducing cost per acquisition or enrollment, increasing engagement, increasing retention, ultimately improving health outcomes, they can attribute it to the AI that they’re leveraging or the productized data science capabilities.

And what we do is we provide them full downside protection. So it’s an ROI driven model. And I think without that, and it’s maybe it’s a little bit different than some of the classic SaaS marketing or MarTech Solutions that are out there that are seat based or purely volume-based. We believe that there’s a value-based approach that can be applied to AI using some older approaches like control groups.

John Farkas:

And that’s a really interesting point, and probably a pretty good challenge to a lot of the folks listening here. What I know about, that’s true right now in this space is, if there’s not a very clear line to value, nobody’s getting much attention. And so your line to value has to be really apparent. And the clearer you can make that and the farther you’re willing to stand behind it, the more likely it is you’re going to have a shot at getting a hearing and finding your way into equation. Because if it’s not apparent, and if it’s not in their critical path right now, the opportunity to get in is only going to be if there’s a very clear line to value. And so I’m curious, Spencer, how you guys are, what you’re doing to position that on the front end in some of these conversations. Because that’s, I’m guessing you’re seeing that the reason you’ve arrived there is because you’ve recognized it by necessity that it better be there.

Spencer Honeyman:

It’s a measure what matters. If you read the book by John Doerr, it’s along that philosophy. And what I would say is, so we provide service across the member life cycle. We have V-Acquire, which is paid media oriented that you would see in the report, and you have V-Engage, which is communication oriented, which can be enrollment, like for an email or a phone or a text through your existing CRM and care management tools. And I think to what you’re alluding to, John, we’ve tried to find the most low friction way to demonstrate value and validate the core KPIs. For V-Acquire, it’s reducing cost per acquisition, for V-Engage, it’s driving enrollment, engagement, retention and lifetime value.

And so I’ll give you a couple examples. If you are the CMO I’m talking to and I’m asking those questions of like, “Hey, if we reduced your cost per acquisition on the paid media spend for your primary care, your urgent care, your dental implants, whatever your service line may be, how do we prove it? We’ll say, “Hey, if you have 10 locations, why don’t you apply our capabilities to four of them and let’s measure the cost per acquisition baseline and we’ll set a threshold for what success looks like.” And we’ll already know post-test, say it’s 90 days, what does a 10% reduction in cost per acquisition look like? That would be one way to do it more on the paid media acquisition side.

The other thing we would do would be, let’s say it’s on the communication side and you don’t need to segment it by locations or geography or anything like that. We can say, “Hey, let’s run a test.” Let’s say you’re a virtual care program and you’ve got a hundred thousand patients. Let’s apply this to 10, 20, 80% of them, have that control group, and let’s prove that we can get more of the right people enrolled in the same period of time using with higher utilization and stay in it for longer. And so we run these tests where we try and provide different ways to leverage our productized capabilities that, although we don’t replace your CRM or your marketing team or your tools, we sit in between the data and those tools, channels and care management platforms. We try and make it low friction to test the capability, demonstrate the value, and then you have, I think, full alignment across all key stakeholders, including the CFO who might be the one writing those checks longer term, that there’s attributable value.

John Farkas:

So it’s a value and it’s your willingness to dig in there, actually elbow to elbow and show them that you can make it happen with the context of a control group and then land and expand, right? That’s the opportunity.

Spencer Honeyman:

Exactly. What we’ve also learned, because I want to share the good and the bad of the audience is, we haven’t, and for the last seven years, since year one, we don’t do anything for free. So what we’ve found is to your point on being elbow to elbow with the customer, we don’t do free pilots. The customer needs to have skin in the game, as well.

John Farkas:

Sure.

Spencer Honeyman:

And they need to be bought in enough on AI in general, and hopefully Vi or whoever they’re working with as a technology or solutions partner that they’re willing to invest their time and dollars. And if they’re not willing to put some skin in the game, the chances for success, it takes two to tango.

John Farkas:

Absolutely.

Spencer Honeyman:

And so that’s another key learning

Closing Question

John Farkas:

And an important one, too, because if they don’t have anything invested, they don’t have anything invested. So they’re interested in successes much less. So talk about how if people are wanting to see the report that you guys put together, how can they find it?

Spencer Honeyman:

For sure. So our website is VI.co, which is V-I-dot-C-O, no M. They can go, you’ll actually get hit, I believe at the top of the website with a banner, do you want to download the state of AI healthcare report? And I think you enter your name, your email, and you can download the PDF or have it emailed to you. And we go through, obviously the most biased person given my role at V, but I think the report is both technology, marketing, product, and business centric. But we try, and like I tried to do today, get into specific examples without providing any information that one of our customers wouldn’t want of what’s working, what’s not, yet, matured, what results are out there in a more tangible way. So hopefully for those listening, it’s a way to get some practical learnings beyond the data science, if you will.

John Farkas:

Yeah, that’s awesome. I think that one of the things we talk about a lot in this arena is the importance as we’re looking at new technology, as we’re looking at what is possible now, the importance of helping people know how to think about that, not just telling them what to think, but the importance of helping people into how to think about the world in light of what’s now possible. And I heard you talk a lot about that, Spencer, in the context of how you’re presenting what you guys are doing and saying, what if we could do X, reduce X by X? What would that mean? And letting that’s giving them something they can relate to that could show them what technology might have the opportunity to help them into, and helping people on that bridge is super critical no matter what you’re doing.

And then being willing to dig in there with them and ensure that they’re going to see that kind of success, not just handing them the tech and say, gosh, I hope it works, but getting in there with them and doing what needs to be done to see that it actually delivers on the promise. It’s important stuff.

Spencer Honeyman, thank you very much for joining us today. I appreciate your time and the insight that you’ve given us and look forward to hearing more about what you guys are doing at V.

Spencer Honeyman:

I appreciate it, John. Thank you for the time and having me.

Outro:

Healthcare Market Matrix is a Ratio original podcast. If you enjoyed today’s episode, then jump over to healthcaremarketmatrix.com and subscribe. And we’d really appreciate your support in the form of a five-star rating on your favorite podcast platform. It does make a difference.

Also, while you’re there, you can become a part of the Healthcare Market Matrix community, and get access to courses and content that’s created just for you by signing up for Insights Squared, a monthly newsletter dedicated to bringing you the latest health tech marketing insights right to your inbox. Ratio is an award-winning marketing agency headquartered in B, Nashville, Tennessee. We operate at the intersection of brand and growth marketing to equip companies with strategies to create meaningful connections with the healthcare market, and ultimately, drive growth. Want to know more? Go to Goratio.com. That’s G-O-R-A-T-I-O.com. And we’ll see you at noon central next week for an all new episode from our team at Ratio Studios. Stay healthy.

Introduction to Spencer Honeyman

John Farkas:

Well, greetings everyone, and welcome to Healthcare Market Matrix. And today we are going to be diving into the wild world of artificial intelligence. And I know everybody is diving into the wild world of artificial intelligence right now, so why not us? But here’s what I want you all to know. I was just at the VIVE conference, I was at HIMSS, I know that every panel discussion or every other panel discussion or maybe 60 something, 70 something percent of the panel discussions at both those conferences had the word AI or the acronym AI in the title, and every booth, practically, had it represented in some way, shape, or form.

So it’s already achieving some level of ubiquity, but does that mean it’s really achieving some level of ubiquity? What’s really being done? How is it really being used? And I think the answer to that is, it’s certainly not quite yet. There’s a whole lot of gaps to close, a whole lot of opportunities to realize. And in a lot of ways we are really just at the dawn. It’s a very big topic and the topic of a lot of conversation right now.

And we are here today with Spencer Honeyman, who’s the Chief Revenue Officer of Vi Technologies, and they’re using artificial intelligence to enhance the outcomes and financial returns of healthcare organizations. And what they’re doing is leveraging AI, a suite of AI powered really, they productize some strong data science to help really optimize the customer lifecycle engagement in healthcare. And I’m probably, you’re going to do a much better job of unpacking that for us, Spencer. But these guys have just put a report together that’s basically outlining the state of affairs related to AI right now in healthcare and pharma. And so we’re going to dive into that today and look at some of what that is. But before we jump in, first of all, Spencer, welcome, and I would love for you to tell us a little bit about your backdrop, what brought you to this moment, and what’s got you here.

Spencer Honeyman:

For sure, John, and thank you for having me. Excited to chat with you and be able to hopefully interact with this audience. My background is I am a startup guy. I don’t know if it was intentional, but this is my third startup at the intersection of health and technology. I’d say both personally and professionally passionate about how can we use innovation to help improve health outcomes, ultimately, and help large businesses by doing that capture some of that value. And so prior to V, I was leading commercial growth for two startups, one in the AR/VR space, which was really interesting and is another evolving landscape.

John Farkas:

Absolutely.

Spencer Honeyman:

And prior to that, was in the hardware software was a company that used what’s called near infrared spectroscopy, or light through the skin to measure biometrics and different things going on in a person’s body. So I won’t fully nerd out on that front since we’re going to nerd out on the AI side, but I’ve spent the last five years with Vi and the company’s been around for eight years. And before ChatGPT was a dinner table conversation, we’ve been looking at how to help large health organizations better utilize their data and create configurable capabilities that technologists and marketers at those companies can use to improve, basically, the key KPIs across the member life cycle and ultimately to help improve health outcomes.

A Deep Dive into Vi’s Recent Report

John Farkas:

Awesome. So help me into, so you guys put a report together here recently.

Spencer Honeyman:

Yes.

John Farkas:

If you were to summarize what that report, what the mission of that report is, how would you characterize it?

Spencer Honeyman:

Yeah, so I think what’s a little bit unique about Vi is we cover healthcare, clinical healthcare, I think some of the largest national and regional health plans and disease and care management with high risk, high cost populations to pharma, whether it’s rare disease or it’s a broader vaccine, to wellness, where we’ll work with a direct to consumer membership or subscription service for fitness, nutrition, mental health. And so we’re looking, we say health, we look across that whole spectrum.

And I think what we try to do internally, and we shared with the world is, where do we think AI is at today? And I think we start by I think picking apart what does AI really mean? And I think there’s different pieces to AI and what’s here today, how can you use it, what is not here, yet? What’s the pace it’s moving at? And then what are the trends we’re seeing in the space? I think what’s interesting, to your point of the AI being slapped on everyone’s booth at the big conferences is, if you look at AI projects or data science projects, still, over 80% of them fail their business objectives. And so I think there’s rapid pace of change. We’ve obviously had different experiences than that 80%, but I think what it means and how it’s used, that’s the goal of this report, to share some wins, some scars, and what we see coming.

Exploring AI Opportunities in Healthcare

John Farkas:

So in the context of what you guys report, as much as it’s dominating the headlines, like we just talked about, it’s everybody, it’s everywhere. The implementation has not been that widespread and what you showed us, 35% of companies have successfully incorporated AI commercially, and like 20% of healthcare organizations have adopted some form of AI. So what do you see continuing to shape the landscape of healthcare and other sectors in the coming years? And what are some of those AI, those opportunities look like for AI right now? How are you looking at that horizon?

Spencer Honeyman:

For sure. So I think one thing we see is, AI is not a thing that you’re just going to be able to plug in across every industry. I think that’s one thing we would say. So I think being vertically focused on using technology, whether it’s AI or otherwise, especially in health, where behavior change is such a hard problem, is critical. And that’s why we choose to focus on that space and we think others will need to, to deliver effective solutions. And I would say we pick apart AI into three different buckets. And this is in the report more eloquently than I’m going to put it verbally, but I’ll do my best.

So I think the first piece is predictive analytics, which is not new, which is how do you use de-identified historical data on demographics, psychographics, different engagement activities, different health populations to predict, in our case it’s different engagement, acquisition engagement or retention patterns in the future, whether it’s a diabetes management program, it’s a health clinic or it’s a pharma vaccine or clinical trial. And so we believe predictive analytics is here and it’s being leveraged and it gets locked into that. I think where we would then move to machine learning would be, well, if predictive analytics helps you use historical data to predict future behavior, well if I know someone is going to disengage from my program, what do I do about it?

And that’s where you can test interventions, whether it’s a call, an email, a text, it’s a reward, it’s an offer, it’s showing up at someone’s doorstep. And what machine learning’s really good at is basically, A-B testing on steroids. How do you test what intervention strategy on what channel at what time works best for different groups of people, and even people on a one-to-one level. And that learning is a lot more sophisticated, faster, and allows for personalization. And so we would say that is here and we’ve seen it in the market very successfully implemented through our customers and others not using our products and services.

And lastly on AI, I think that in our lexicon, that would mean more automation and more generative AI where you put an input and you expect them to generate the email template, the advertisement that would go on Google or Meta. And so what we do at Vi is we’re not in the generative AI space now. What we do is we provide frameworks to marketers and product and technologists teams at our customers to use predictive analytics, some data that we license on consumer behavior, and machine learning to optimize their content, their email script, what the nurses should say on the phone, what the automated tech should say. And I think what we believe generative AI is here and coming, but that fully automated robot, like Terminator, experience we think is a little bit far out. So we’re excited about it and it’s moving fast.

John Farkas:

Let’s hope the Terminator experience is more than just far out.

Spencer Honeyman:

Yes.

How AI Has Alleviated Administrative Tasks

John Farkas:

Let’s hope it’s not going to happen that way. Yeah, so we’ve got three decent sized buckets here. It’s the use of how do we get over some of the administrative things, which everybody agrees uniformly. Some of those, the busy work administrative hurdles are some of the low hanging fruit in this realm, and then we have the member experience space, and then we have the predictive analytics space.

If knowing that, let’s dive into the administrative space to start with here. And looking at some of the administrative tasks like scheduling, and claims processing, data entries, those alleviating some of that workload, and for healthcare professionals is some obvious opportunities, so they can focus more on engaging their patients and actually providing the care. So can you provide some examples where AI driven solutions have been successful in alleviating some of those things for healthcare professionals and really doing the work of improving efficiency for some of those workflows?

Spencer Honeyman:

For sure. I think for the vast, vast majority of healthcare professionals, there’s never enough time in the day. And so I think operational efficiency is obviously critical. And so I think a couple of examples, I think for both I’m going to share here, it starts with having data that is in an infrastructure that can access both historical and ongoing data to be able to help these things be automated in a way that works. And so it’s cliche and it’s been a garbage in, garbage out since the days of big data, etc. But I think it does still apply and you need to have a way for the AI or the technology to learn.

I think two really interesting examples. So we have a customer, they’re a very large health system, and they had a lifestyle coaching service for those with pre-diabetes or early stage diabetes. And what they found was the health outcomes were most linked to the compatibility of the patient with their provider, and so, because this was an ongoing coaching process. And so what we found was really interesting was A, how do you match, how do you look at the good matches and “the bad matches” or ones that weren’t a great fit. So you’re able to, in an automated way, figure out, as that patient comes into the system, who would be the right health professional for them to work with. And then what was really interesting was we started to see patterns and we were able to automate some things like scheduling the next appointment at a time that works for that health professional, in general, and matching it to what works with the person receiving the care, the patient.

And what we found was not only better results, but also we were able to increase bandwidth and productivity for that healthcare professional, the number of patients they could serve in this diabetes lifestyle management program, we were able to increase it over 35%, which is obviously transformational, in terms of the impact that organization could have on the population.

Employing AI to Increase Patient Engagement

John Farkas:

Yeah, no doubt. And we’re seeing lots of different manifestations similar to that across the ecosystem with a variety of different solutions. When you are looking at, if we were looking at things related to member engagement and enrollment, engaging members is obviously one of the big challenges that is existing right now. It’s so hard that Walmart can’t even do it, apparently. But as we are looking at the advancement of AI and how what kind of innovation, innovative solutions can come to the table in that regard, are there specific populations or healthcare scenarios where those kinds of interventions are really showing some promise or look really promising on the horizon as far as what we anticipate being able to happen?

Spencer Honeyman:

Yeah, I think what has remained true and what will continue to remain true is there’s this unfortunate correlation between certain health populations that have high chronic risk factors, usually more than one of them, and how hard it is to engage those in the care they need and keep them in that care to ultimately improve health outcomes. But I think what we have seen is that you’re able to create the uplift, in terms of the engagement level of those people if you apply different technology and AI frameworks.

And so to make that a little bit more practical and give you some examples, so we work have health plan customers and they will have certain partners or internal programs that go across, let’s say chronic disease management, so MSK diabetes, kidney disease management, behavioral health and more, and I think we’ve seen a couple things be really effective. So one is I think in a HIPAA-compliant way, there are ways to leverage third party data sets that are available. We do some of this at V, as well, to enrich the data health organization may have, but not related to claims or Rx data. So for that profile of patient that you’re looking to engage, where do they go? What retail stores? What restaurants? Do they go to a health club? What do they search online? What different behaviors do they have when it comes to what they purchase, things of that nature. And you’re able to create more of a 360 view versus just the traditional healthcare only data view. And I’ll give some examples shortly of how that can be applied.

And then I think what we believe is the force multiplier, or what we would call optimization frameworks. So I think we’ve seen from marketers, specifically around engagement, I don’t think going to tell you, hey, here’s the perfect email script you should send to this person, and magic happens. But I think there’s now frameworks where you can leverage the data at your disposal, different third party outside data that is not healthcare related to rapidly test and iterate. So I’ll force you and the audience to nerd out with me very shortly, John.

But I think if you think of AB testing historically, let’s say we have two email templates and we’re trying to get this individual to enroll or engage in a certain program or to sign up for a PCP, whatever the use case would be, and we test A verse B and either A or B wins. I think the difference now with machine learning and AI is you can have A, B, C, D, E, F, G, and in the past where B may have lost to A, B may “lose”, but it might be really good for 3% of the population that’s Medicare in the Northeast that has X, Y, and Z related to them. And we don’t need to throw it out, but we can serve up the right content to the right individual on the right channel. And those subtle tweaks of what works for different people eventually, either in much smaller segments versus a one size fits all or even at an individualized level has huge impacts on whether you’re able to even contact or communicate with your target audience or whether you’re actually able to engage them.

John Farkas:

Yeah, very cool. What are you seeing as far as how eager or open some of the health systems are to making that happen? It’s one thing to talk about, it’s one thing to say we could do this. It’s another thing to actually get the buy-in and permissions and all the things necessary to actually see it through.

Spencer Honeyman:

I think it can vary and it’s certainly, I think, a challenge that the healthcare industry has versus maybe traditional consumer retail of obviously we’re handling very sensitive data. So I think there’s some approaches that have worked better than others. So for us, we’re not in, we don’t predict medical risk, we don’t make care recommendations, what we tell our customers is we can help you find more of your target population, get them in doing more of what you want them to do for longer. And if your care and what you do for them leads to improved health outcomes, that’s our mission and we want to help you do that better. And so because of that, we don’t, in most cases, we don’t need to take in PII, we don’t need to know it’s Bob Smith, I don’t need to know what their A1C was. Maybe we know if they took a glucose reading, and it’s just member 1, 2, 3, 4. And so I think there’s ways to de-identify the data and use more engagement-centric data, specifically for marketing. I think that’s one thing.

I also think second, where we’ve seen a huge change is around value-based care. And so it’s something that’s obviously not new and I’m personally passionate about, but I think the shift in incentives is pretty massive to where organizations are now able to leverage paid media channels. And I’ll give you an example. So let’s say in the past, if I was providing a certain service, let’s say primary care, and my employer, I sell to large ASOs to self-insured employers, and they were covering the cost, in the past where it may have been 2 cents per employee per month, whether they use the service or not, now it’s $250 per engaged member per month.

And while that creates the right incentives, the other thing it does is, hey, now if I can leverage de-identified data to serve up an advertisement on where that individual is today on Google, or Instagram, or Facebook, or Hulu, or wherever it is, I can spend $100, or 200, or $500 to acquire that patient, in the marketing sense, on paid media in addition to traditional non-paid communication channels. And so what we’re seeing customers do is start leveraging both communications and media to meet these audiences where they’re at because the financial economics of some of these value-based care models align.

An AI Success Story

John Farkas:

So Spencer, let’s take the value-based care scenario. What’s a good example of something you’ve actually seen in the market where that has come into play? How has that been realized?

Spencer Honeyman:

Yeah, we saw, we have a pretty transformative case study and this always, this is not a one size fits all, I think that’s for sure with all AI, but we have a-

John Farkas:

That’s what I was going to say, that’s the state of affairs, right?

Spencer Honeyman:

That is for sure.

John Farkas:

Or at least it should be.

Spencer Honeyman:

I think to get into that success using any of these buckets of AI, I would fully agree. I think one example that we thought really interesting was pretty transformative for one of our customers is their regional behavioral health clinic, and they focus on children with rare and pretty severe forms of autism. And they offer this as a covered benefit to large employers and health plans for, usually for their children. So it’s like think ages 2 to 12, or who are the children that have this autism?

And what they really struggled with in the past was they were using communications, and it was really hard to engage the parents or caregivers of these children, explain to them what this is intensive program that they offer. It was multiple times a week for an extended period of time, could provide to them that’s different, maybe, than the care they’re receiving from their primary care or they saw in general, and it was hard to reach them. And also, a lot of times they don’t have permission from that employer or health plan to send the email they want to, when they want to, how they want to. And so using the value-based care example with this shift to only getting paid based on engagement and health outcomes, they were able to start using media.

And so what we did was, with them, their marketing team was able to look at a bunch of analytics of, okay, who are the parents of these children and what do they do? So what are their demographics, age, gender, location, what does their mobility look like? Do they shop at different stores? Do they purchase certain things for their children? That would be very non-obvious, and I obviously can’t go into detail here, but nothing to do with what you would consider a medical device. We’re talking consumer product purchases, that if you have a child with rare form of autism that this would be applicable for, would make you a lot more likely to be in this target audience. What do people search? And again, it can’t be health related, but it could be X, y or Z, Again, without going into too much detail. And do you follow key digital opinion leaders or join advocacy groups?

And so what we helped them do was basically create these segments and use data sets that we have access to, to be, call it 10 to 20 times more precise with who are the mobile IDs and IP addresses without ever identifying these individuals that look like the people that are the caregivers of the parents of your target population. And then they were able to deploy media to those individuals across social search, connected TV, they did some direct mail and they were able to reduce their costs for acquisition over 35%, and their conversion rate went up over 50%. So it was very quality set of leads, which was transformative to them of how do they get the right people in the door, and ultimately provide care in the way that they do, but more economically.

The Awareness vs. The Adoption of AI

John Farkas:

So Spencer, I know that there’s awareness of AI and then there’s adoption of AI, which are two different things. And I know that you, being in the position you’re in within your organization, you are in the conversations where people are confronting some of the hesitations, the concerns, the objections, the fears, whatever those might be. Talk to us a little bit about what you’re seeing in those conversations. What are some of the hurdles that you’re facing that are, I know, if I know one thing, if you’ve sold in the one health system, you’ve sold in the one health system, so there’s any number and they’re varied. But what are some of the common themes that you hear as far as objections or hard, things that make it difficult for them to consider the kind of innovation that we’re talking about here?

Spencer Honeyman:

Yeah, so I think for the most part we’re talking to leadership within these organizations and marketing, product, technology, and data. So it can be a broad group. I think the biggest thing is, is this actually working? And if so, how much and how do I attribute it to this versus all the various factors that can go on in the world. And so what we’ve seen work well on our end is, let’s say we’re speaking to a marketer, and we can go back to the example I gave on, let’s say I’m using paid media acquisition or optimizing communications and email marketing to the right person at the right time through their existing CRM and tools, we would say, “Hey, what would it look like if we could increase your enrollment and engagement 20% or your retention in this program 20%? What would that mean to you as a business, as an organization?”

John Farkas:

Classic challenger sale tactic, right?

Spencer Honeyman:

There you go. And you say so. And if they don’t have an answer, then I think they’ll be like, “Okay, that’s interesting. Let me come back to you and think about it.” And what our model is, and this is after years of pain, and there’s plenty of models that can work here is what we do is we sign long-term agreements with these customers. Usually, let’s say it’s a CMO, that’s our partner that’s actually signing this agreement and we say, let’s create a back of the napkin of what this value could look like.

And everything we do has a control group. So if the world changes, your marketing content changes, you expand, you retract, things go on that are not relevant to what we’re touching, we’ve got a holdout group that it will be a consistent baseline. Not just during a pilot, but always. A, because it’s how the machine learning learns. We’ll still predict on that control group or the holdout group, if you will, but you’re able to attribute the value and the learnings that you’re seeing, whether it’s reducing cost per acquisition or enrollment, increasing engagement, increasing retention, ultimately improving health outcomes, they can attribute it to the AI that they’re leveraging or the productized data science capabilities.

And what we do is we provide them full downside protection. So it’s an ROI driven model. And I think without that, and it’s maybe it’s a little bit different than some of the classic SaaS marketing or MarTech Solutions that are out there that are seat based or purely volume-based. We believe that there’s a value-based approach that can be applied to AI using some older approaches like control groups.

John Farkas:

And that’s a really interesting point, and probably a pretty good challenge to a lot of the folks listening here. What I know about, that’s true right now in this space is, if there’s not a very clear line to value, nobody’s getting much attention. And so your line to value has to be really apparent. And the clearer you can make that and the farther you’re willing to stand behind it, the more likely it is you’re going to have a shot at getting a hearing and finding your way into equation. Because if it’s not apparent, and if it’s not in their critical path right now, the opportunity to get in is only going to be if there’s a very clear line to value. And so I’m curious, Spencer, how you guys are, what you’re doing to position that on the front end in some of these conversations. Because that’s, I’m guessing you’re seeing that the reason you’ve arrived there is because you’ve recognized it by necessity that it better be there.

Spencer Honeyman:

It’s a measure what matters. If you read the book by John Doerr, it’s along that philosophy. And what I would say is, so we provide service across the member life cycle. We have V-Acquire, which is paid media oriented that you would see in the report, and you have V-Engage, which is communication oriented, which can be enrollment, like for an email or a phone or a text through your existing CRM and care management tools. And I think to what you’re alluding to, John, we’ve tried to find the most low friction way to demonstrate value and validate the core KPIs. For V-Acquire, it’s reducing cost per acquisition, for V-Engage, it’s driving enrollment, engagement, retention and lifetime value.

And so I’ll give you a couple examples. If you are the CMO I’m talking to and I’m asking those questions of like, “Hey, if we reduced your cost per acquisition on the paid media spend for your primary care, your urgent care, your dental implants, whatever your service line may be, how do we prove it? We’ll say, “Hey, if you have 10 locations, why don’t you apply our capabilities to four of them and let’s measure the cost per acquisition baseline and we’ll set a threshold for what success looks like.” And we’ll already know post-test, say it’s 90 days, what does a 10% reduction in cost per acquisition look like? That would be one way to do it more on the paid media acquisition side.

The other thing we would do would be, let’s say it’s on the communication side and you don’t need to segment it by locations or geography or anything like that. We can say, “Hey, let’s run a test.” Let’s say you’re a virtual care program and you’ve got a hundred thousand patients. Let’s apply this to 10, 20, 80% of them, have that control group, and let’s prove that we can get more of the right people enrolled in the same period of time using with higher utilization and stay in it for longer. And so we run these tests where we try and provide different ways to leverage our productized capabilities that, although we don’t replace your CRM or your marketing team or your tools, we sit in between the data and those tools, channels and care management platforms. We try and make it low friction to test the capability, demonstrate the value, and then you have, I think, full alignment across all key stakeholders, including the CFO who might be the one writing those checks longer term, that there’s attributable value.

John Farkas:

So it’s a value and it’s your willingness to dig in there, actually elbow to elbow and show them that you can make it happen with the context of a control group and then land and expand, right? That’s the opportunity.

Spencer Honeyman:

Exactly. What we’ve also learned, because I want to share the good and the bad of the audience is, we haven’t, and for the last seven years, since year one, we don’t do anything for free. So what we’ve found is to your point on being elbow to elbow with the customer, we don’t do free pilots. The customer needs to have skin in the game, as well.

John Farkas:

Sure.

Spencer Honeyman:

And they need to be bought in enough on AI in general, and hopefully Vi or whoever they’re working with as a technology or solutions partner that they’re willing to invest their time and dollars. And if they’re not willing to put some skin in the game, the chances for success, it takes two to tango.

John Farkas:

Absolutely.

Spencer Honeyman:

And so that’s another key learning

Closing Question

John Farkas:

And an important one, too, because if they don’t have anything invested, they don’t have anything invested. So they’re interested in successes much less. So talk about how if people are wanting to see the report that you guys put together, how can they find it?

Spencer Honeyman:

For sure. So our website is VI.co, which is V-I-dot-C-O, no M. They can go, you’ll actually get hit, I believe at the top of the website with a banner, do you want to download the state of AI healthcare report? And I think you enter your name, your email, and you can download the PDF or have it emailed to you. And we go through, obviously the most biased person given my role at V, but I think the report is both technology, marketing, product, and business centric. But we try, and like I tried to do today, get into specific examples without providing any information that one of our customers wouldn’t want of what’s working, what’s not, yet, matured, what results are out there in a more tangible way. So hopefully for those listening, it’s a way to get some practical learnings beyond the data science, if you will.

John Farkas:

Yeah, that’s awesome. I think that one of the things we talk about a lot in this arena is the importance as we’re looking at new technology, as we’re looking at what is possible now, the importance of helping people know how to think about that, not just telling them what to think, but the importance of helping people into how to think about the world in light of what’s now possible. And I heard you talk a lot about that, Spencer, in the context of how you’re presenting what you guys are doing and saying, what if we could do X, reduce X by X? What would that mean? And letting that’s giving them something they can relate to that could show them what technology might have the opportunity to help them into, and helping people on that bridge is super critical no matter what you’re doing.

And then being willing to dig in there with them and ensure that they’re going to see that kind of success, not just handing them the tech and say, gosh, I hope it works, but getting in there with them and doing what needs to be done to see that it actually delivers on the promise. It’s important stuff.

Spencer Honeyman, thank you very much for joining us today. I appreciate your time and the insight that you’ve given us and look forward to hearing more about what you guys are doing at V.

Spencer Honeyman:

I appreciate it, John. Thank you for the time and having me.

Outro:

Healthcare Market Matrix is a Ratio original podcast. If you enjoyed today’s episode, then jump over to healthcaremarketmatrix.com and subscribe. And we’d really appreciate your support in the form of a five-star rating on your favorite podcast platform. It does make a difference.

Also, while you’re there, you can become a part of the Healthcare Market Matrix community, and get access to courses and content that’s created just for you by signing up for Insights Squared, a monthly newsletter dedicated to bringing you the latest health tech marketing insights right to your inbox. Ratio is an award-winning marketing agency headquartered in B, Nashville, Tennessee. We operate at the intersection of brand and growth marketing to equip companies with strategies to create meaningful connections with the healthcare market, and ultimately, drive growth. Want to know more? Go to Goratio.com. That’s G-O-R-A-T-I-O.com. And we’ll see you at noon central next week for an all new episode from our team at Ratio Studios. Stay healthy.