Why Healthcare AI Adoption Stalls in Real Workflows

From the episode captions, lightly cleaned. Speakers are not labeled, and the text may contain small errors. Time links play the episode from that moment.

0:00 What are the challenges that you're seeing today I in today's marketplace in health care? 99.9% of people who leave health care schools um are passionate about helping sick people. Like they want to get in front of people who are not feeling well to make them better. Some of the highest impact that we can make in health care is looking not at necessarily um their traditional divides in health care delivery, but looking at the challenges that the rural and the um the areas where health care entities are very distant and far between each other.

One of my big passions is trying to make sure that everyone realizes that um for every friction point that we think we can design away with AI and technology that isn't designed away, um it's going to fall onto a person to absorb. Hey there. I'm Terry Coutsolioutsos, founder of RevRx AI. We remove friction and architect revenue for medtech and consumer health practices. Welcome to the Zero Friction Podcast. This is Terry Coutsolioutsos and once again I have an extraordinary guest with me today. I have Vishal Amin, who is the founder of and owner of Health E Tech Consulting. Vishal has quite an extensive background in the pharmacy and informatics arena. He helps companies understand those data points, creates opportunities for efficiencies, and so um I'm going to let Vishal talk a little bit about himself, his background, his experience, and then we're going to lean into his consulting company on what he's helping other companies transform in what they do in delivering health care today. So Vishal, welcome to be on the

Show. Appreciate it. Yeah, happy to be here. Tell me about yourself. Yeah, sure. So my name is Vishal. I'm a pharmacist by background. I have drug information residency experience, and primarily I work in clinical informatics, which is the space where we help technology hit the ground and let end users and patients use the things that we build in order to get higher quality health care. So, I had a chance to work at CVS Health for a number of years, primarily building things like clinical engines and user interface elements in our workflow to fill billions of prescriptions over the course of a decade. I had a chance to work between CVS Health and Epic for a number of years doing implementations across the enterprise. I think a key area of focus that I work in, and which I think we'll get to, is human factors engineering, which is using a set of skills and tools to make sure what we build can be maximally useful for end users. And, like you said, I have a

2:39 Consulting firm now where I work on my own with a variety of different companies on difficult and unaddressed health care technology problems. So, I'm really excited to be here and I appreciate you having me on. That's great, Vishal. And it sounds like having, you know, you've got quite an extensive background. You have a clinical background, you have an informatics and a technical background, and, you know, now we're getting into the human factor component. So, what I find most interesting is combining all these things together. What are you and your consulting firm doing to help change the way care is being delivered today? Yeah, the the key differentiator that I think we can bring to the marketplace is human factors engineering is a um pretty well-defined applied science field of study. It's used in a lot of different domains, including health care, but being able to bring it from a a pharmacist background into the technology space, it kind of allows me to look at and analyze workflows more

Deeply with a unique set of tools and figure out the right technology to put into place to address those problems, whether we're talking about at the beginning or designing the system, in the middle where we're encountering problems and refining the way that we're getting our product into the marketplace, and then once we've launched it, being able to prepare our end users to use it to the top of its ability, and then continually refine it over time. What are the typical friction points that you're helping these companies solve for today? Yeah, I I usually categorize them in two large buckets. There are the first set of problems, which is they know that something's not working right, but they're not exactly sure how they're going to solve it. In many cases where this might be happening, it could be just a couple of people who have identified or voiced the problem, and then the whole solution is built directly on those people's feedback. In those sorts of cases where I can help is to take a look at the workflow in a

Little bit more detail using some of those human factors tools. One example is doing a detailed analysis on the way that the end users might be thinking and acting and organizing their tasks, not only for the one problem that they encountered, but for all of them. You'd be surprised how many times a detailed workflow diagram can save you hours and hours of headache later on when you're looking to deploy out these tools. And then another set of problems, and these typically happen at larger organizations, is getting items into production, right? Like getting them in front of your patients or getting them in front of your end users can sometimes be a gauntlet. You know, there are a lot of different teams in large organizations that might have a say, or if you haven't thought about it from the regulatory space or the legal space or the patient safety space, it can kind of kill products and projects dead in the water sometimes. So, using my experience bringing these sorts of products to

5:34 Market for really large organizations, I can usually put a plan together that addresses a lot of these boxes and make tweaks on the fly to the to the plan so that we can actually launch these and get these into the hands of people rather than letting them hit a block that we can't get across. So, tell me about something, you know, you talked about you own one of the largest Epic, you know, implementations ever, right? And so, tell me what it's like to work with a massive system that is serving hundreds and thousands of stakeholders.

And what is that like in terms of, you know, creating an end delivery item for the customer? Yeah, when I when I was working at CVS Health and uh partnering with Epic at the time. But, most people don't think of CVS Health, the retail pharmacy player, as like a large consumer of Epic, which is a electronic medical record typically used in the inpatient setting. But, Epic is a very flexible tool. It has a variety of different modules for different use cases. And we consumed a lot of their non-inpatient, non-hospital-focused modules in a variety of different ways.

We used modules in our pharmacies themselves to deliver care, especially during COVID times. Uh we had health plan partners and telephonic care management partners who were using the tools to keep their work organized and engage patients to address problems with social determinants of health. We had outpatient clinics, MinuteClinic, who was a large consumer of it as well. And so, the the real learning experience there was to see how do you apply such a flexible framework to like a variety of problems. I got a chance to work with different parts of the organization who were looking at different platform solutions, trying to figure out from a cost-effectiveness perspective would a vendor product like Epic fit their needs? Because with large organizations who have the resources to do so, it's always a decision that they get to make.

Are we going to build it? Are we going to buy it? And each one of those decisions comes with a set of tradeoffs, right? If you build it, you're going to get a large amount of flexibility. You're going to be able to customize it to the way you want to do business, extract value in the way that you believe you can position yourself in the marketplace, but you're going to need the infrastructure to do it, right? You're not only going to need the hardware, you're going to need the people, not only to build it the first time, but to keep it going for the long run. Versus if you go with the vendor system, you are a little more constrained as to what the tool can do, but you're getting a set of expertise infused into your system, right? You don't need to refine it over time. So, whether you do one or the other is an important decision to make up front, and I was able to work with the business stakeholders who wanted a new platform

8:19 Or were thinking about one and figure out, all right, what is it exactly what you need, how much might that cost in the long term or the short term, what are the trade-offs if you wanted to build your own system or migrating off something legacy. Um, so it was a great learning experience in a variety of different use cases across healthcare, cuz at the time the enterprise of CVS Health was looking at it in a variety of different ways. Yeah, systems like that, and I've I've had the opportunity in my career as well in terms of not maybe an Epic EMR system, but I I did deploy and and design a CRM system for seven different sales divisions inside of one company. So, it's I do understand the balance between buy it or build it. And what was the final straw in terms of how CVS decided the buy versus the build component?

Yeah, the the ROI models and the three-to-five-year projections of how highly they might be consuming that part of healthcare was a was a big factor. So, uh certain parts of the company might be using it to directly see patients and document into it. Some parts of the company might be using it almost like a secondary or tertiary system where their primary system was used for things like filling prescriptions in the case of the retail pharmacy, and then their augmented services might have been something like seeing patients for COVID reasons or prescribing hormonal contraceptives or something like that. And so, like the frequency of how often they would use it played a part. Another large piece of it that played a part is the interoperability of it. So, Epic I think is a strong player in the healthcare marketplace because they really thought through how they wanted to bake in interoperability, the sharing of clinical healthcare data amongst healthcare institutions. And parts of the company that were looking at use cases that would highly leverage that

Interoperability found that to be a key piece of the puzzle that they weighed heavily when they were considering should we or shouldn't we use it. What I think was surprising to me that a lot of the use cases didn't really need time to think through was are Epic's capabilities good enough for what we're doing? Cuz actually a lot of them were coming with very mature operations. They had been doing it for quite a while, and either they had an older system or the older system isn't available anymore, or maybe they didn't want to maintain their own system anymore. I do work with other clients in different sized organizations, though, where it's a net new question for them.

10:50 And so, then it becomes a little bit different of an equation, right? But for most of the use cases I was looking at, that didn't play as big a role in in some of those decisions. Got it. You know, I just know how complicated it is, and I know how painful those types of projects are. So, I just, you know, I I commend you for making it through all those things cuz I can't imagine the stakeholder messes that that go through that. But I would also say at the end of the day, you know, if I were sitting in the middle of the healthcare world like a retail a large retail pharmacy does sit, whether it's CVS or other large, you've got to have the same systems that healthcare providers are using in order probably just to make the data flow a little more integrated and less friction from a data perspective to make sure that everything is connecting the right way.

Yeah, absolutely. And that's where the that's kind of the the groundwork where I like to get hands-on and it becomes a lot of fun for me. You know, trying to figure out how do we merge our master patient indexes, you know, how much weight are we going to give an approximate first name match versus a exact last name match and those sorts of things. And And what difference does that make to our outcomes, right? That's some of the the painful parts of working in the interoperable space, making sure data can be used across healthcare institutions, but the fun part of working at scale is once you solve that problem, you solve it for hundreds of thousands, millions, hundreds of millions, you know, the numbers just kind of grow and and give at least me a lot of fulfillment. And that's what makes me really enjoy the work that I I get to do.

Absolutely. Well, let's take a pivot real quick because the world is moving very, very fast. Healthcare is We'll talk about AI. I'll frame it up a little bit. And healthcare is I think the third fastest industry or market segment that's adopting AI today behind fintech and insurance, right? And the insurance industry is tied right into the whole healthcare piece anyhow. So, you can almost combine them and that's they're it's the second largest. So, talk to me a little bit about AI and the concept of AI around the center of healthcare. And what are a couple What are the challenges that you're seeing with AI in today's marketplace in health?

Yeah, I I think that for my experience being hands-on with these sorts of projects, uh And And I guess the first point to make is I do think that most sectors were working with elements that you would consider AI before the like large growth that we saw with the generative AI models and the LLMs. And so, a lot of the the the truth of it now that I think some of the excitement is dying down and we're starting to get back to what I would say more normal times when it comes to assessing these technologies. Now that we've seen if we just implement without a real structure in place, the cost can start to spiral out larger than our ROI can be. The truth is is that a lot of these fundamentals that we looked at for implementing any technology still are true. We need to understand the problem that we're trying to solve at a very core level. We need to have analyzed where we're going to use these different technologies and truly assess the

14:04 Tradeoffs that we're going to have because any new core technology you're implementing is going to change the role of the humans and the users who are in that system and we have to understand is that going to be sustainable and worth it? And how are we going to implement it in a way that is safe. And you know, there's a lot of different facets to safety when we think about AI. And in health care, you have the added factor that you need to make sure that you're HIPAA compliant, that you're taking care of people's health care data in the right way. In addition to making sure that the way that it's being used, the outputs are safe to use in patient interactions, in documentation purposes, and that our end users have confidence in how we're using it. And so, a lot of what used to be true for any technology implementation, I think still holds true in a large part for the new wave of AI

That's being implemented. You you got to understand why you're using it, make sure you have an a strong ROI that you're marching towards, and that'll keep you tethered in how you make the implementation decisions, how you make the tradeoffs, what you do in the next 30 days versus the next four months. And so, you know, a lot of those same guiding principles I'm seeing are where folks are finding success in implementing AI solutions. You know, and there are certain things that get in the way of any types of adoption of any kind of new platform or technology. What are you finding are the biggest friction points getting in the way of more AI adoption today in health care? Yeah, I think a big one for larger organizations is making sure that the whole company has a strategy and that individual projects are fitting underneath its framework. That includes the architecture, the governance, uh the safeguards that you're putting into place. How are you measuring that it's

Meeting security standards? That's a very large one that can be outside of the realm of the project itself, right? That's more of a organizational decision. And the other one, which is a big one, is I think going too big too quickly. It can be very tempting because you can see the power of generative AI in your hand and in your pocket when you have it for home use and you see how like transformative it can be at home, generating recipes or helping you organize travel or a lot of people have it in hand and they see how much it can improve your life. I think one thing that gets in the way a lot of friction is that we want it to do everything on day one and so we want to implement it for everything on day one. And it can't do everything perfectly. And if it can do 20 things at 80%, that's going to kill the implementation much quicker than if you had it do two things and you

16:47 Got it to like 99% effectiveness. And that friction point is typically one that is a leadership to implementation team plus or minus vendor team that I think needs to be navigated and kind of navigated in a, you know, in a informed kind of way and in organized kind of way. You know, it gets back to that human factors problem where, you know, the things I think you're leaning in towards discussing is like the the infrastructure of the performance all need to be there because if it's not, then you lose trust. And when you lose trust, that's when you lose adoption.

And not just lose adoption for the moment, but you might lose adoption in the momentum of what you're trying to create over time as well. And so those are the things that are always been challenging from any technology adoption phase when you have change management teams or transformation teams or whatever you want to call them, right? But being building something that is going to deliver performance is always key, right? And then so the the more it perform where you're saying I'd rather have it do two things at 99% versus 10 things at 80 because those two things at 99% will lead to the third at 99%, which leads to the fourth at 99. And then, you know, you have a confident team implementing these things across the board. You know, the the other thing that you worry about too with AI, and I think whether it's internal or external, but I do remember building and working with some remote patient monitoring technologies probably 20 or 30 years ago before it was really mainstay. And I

Remember turning on some of these technologies in the pacemaker and defibrillator business, and you know, the alert systems were crazy. And I remember like when we first started using them, we really didn't understand the energy that came with all these alerts, and it really sort of fatigued, you know, a lot of clinicians around not wanting to have this information because they were just so overwhelmed by it. So, you have to learn how to titrate that information and titrate those alerts so that you're not scaring people away.

Um, number one. Number two is when they get so much information, it becomes noise, and then even the important stuff gets unnoticed. And so those are things that, you know, any kind of technology adoption probably should have some sort of measurement or sort of, you know, perspective on how to how to tackle those things. Yeah, and I, you know, the this is is area where I've spent a lot of time both learning about through my human factors training. You know, there are models within human factors like signal detection theory that talk about how to properly and mathematically figure out where the signal and the noise are happening when you talk about alert fatigue. But in my experience, I think the systems that are most successful aren't those that go out the door with the right settings, especially if you're operating at scale, because to know what the right setting can be assumes there is one right setting and the one that is right today will be right forever.

19:43 That's just not how clinical practice works. The systems that are successful dealing with alert fatigue are the ones where the team that's maintaining the tool, number one, has like a robust set of levers they can pull to move the different types of alerts around to hide, suppress, elevate. They have different options on how to use the information within the system to move the alerts. And number two is that they've thought through and have a monitoring plan to see how well is it going. Do you know what the signal is and what the noise is? Can you measure it? And then once you've measured it, can you then modify those levers that you put into the system in the first place to tweak it over time in order to maintain a healthy signal and noise relationship. So, alert fatigue is one that takes many forms in clinical decision support. It's maybe the most important, like the most paramount element to a good clinical decision support system.

But what I find it from my experience is that it's an ongoing kind of relationship between a team responsible for its care and the tools that they're using to to effectively alert their users to, you know, drive good outcomes for people. Let's pivot again real quick. I know you're working on some exciting things in the physical, like moving the physical world. Is there anything that, you know, that you can kind of lean in on that a little bit with us and you know, tell us a little bit more about that transformation.

Yeah, yeah, I'm I'm working with a client more on the physical side, which is a completely different side of your brain to flex, which has been a lot of fun. Thinking about how do we operationalize and integrate robotics and the new wave of technology we're seeing over there in order to physically deliver health care in a more effective and efficient way. And so what we've been trying to look at is how health systems and hospitals that are either standalone buildings or maybe they own a campus of buildings that are clustered together. How can we use the emerging wave of robotics and drones in order to enhance their health care offerings such as delivering medications around the different facilities, how do you move things like labs and blood work and IVs, enable them to refocus their personnel on the patient contact and the human contact rather than having to focus on those more logistical problems.

And so I've been I've been working with in the startup space mostly there on how do you build like logistical networks in order to make that happen. Is that like within hospital campus systems or are they actually are you talking like through like the city or like what's sort of the concept behind that? Yeah, the the primary consumer I think would be hospital systems themselves. Those that can take advantage of the technology. So what we're looking at is can we partner drones and the company that I work with builds robotic shuttles. And so you pair them together and you can move payloads from one building to the next building or buildings that might be a few blocks away. This will be really useful in both urban settings where it can be difficult to navigate couriers or cars, but also potentially in rural settings where the buildings might be a little further apart. And the key problem that we're trying to solve right now is organizations that might find value in implementing drone technology. It can be

23:13 Hard to operationalize it cuz you don't always have the ability to send your personnel to the roof or out to a parking lot or somewhere where these drones might land. So, how might we use robotic shuttles to take the payload off the drone and bring it right into the building? Yeah, cuz you have I'm sure you have drones that fly, you have drones that roll, you have all kinds of different, you know, uh drone technologies that can move things from one to the other. So, how I found it extraordinarily exciting just the the concept of building something like that in the healthcare arena is is must be must be filled with great purpose and intent, for sure. Yeah, and there's, you know, there's you can think in the more macro about these sorts of portal hub systems of robotic shuttles that can integrate between a drone and maybe there's a locker at ground level or there's a drive-up kiosk or there's a handoff to a little one of those like little dog drones that walks around and

And, you know, can drop it off at your at your porch. But also, you can make the system a little bit smaller and think about moving things inside of a single building. And so, we're excited to kind of explore all the different applications they might have in healthcare. And it really scratches that same kind of human factors itch for me because these technologies are not going to work ambivalently behind the scenes where you never see them in all cases. There are going to be human-to-machine handoffs and getting to think through like the failure points because if those aren't clean, companies aren't going to want to use this because they're going to be impossible to operationalize. And that's the kind of problems that I like to find, seek out, and and solve. So, it's a different portfolio of work from what I'm typically working on, which is the IT systems themselves on the software side, but you know, really exciting stuff and I think it can

Make like a really large difference in people's lives as well. So. You know, another thing that we had talked about earlier that will make a a huge difference in patients' lives, too, especially for those that have access to health care issues, you know. Um, tough tough to access geographies and talk to me a little bit about the work that you've done in the past regarding, you know, remote areas and, you know, and specialty pharmacy explosion in certain areas and and how how is AI helping that?

Yeah, I think um, you know, I had a I had a mentor of mine who laid this out much better than I'm going to be able to, but the gist of it is that some of the highest impact that we can make in health care is looking not at necessarily their traditional divides in health care delivery, but looking at the challenges that the rural and the the areas where health care entities are very distant and far between each other. And so access deserts are becoming a big focus, I think, in health care across the board. Specialty pharmacy is definitely one component of that, but just any type of care where people have to go a long distance. And I think what pulled my attention into this space is during pharmacy school, I had a chance to work with the Indian Health Services in Arizona and my experience there really changed my perspective is people who would come there for health care at these facilities would sometimes walk

26:27 Two, three, four hours from where they live to receive like half an hour of care. And they were always like so appreciative, very patient, and it just gave you the sense that like their kind of the challenges in their life are just very different than the ones that you might experience in the suburban or urban area. And that the value that you could bring, if you could solve some of the issues that they might have to health care access would be both highly rewarding, fulfilling, and very helpful to them. And so, you know, the the work that we're doing on some of these drone and you know, logistics, robotics technologies, I think are going to have a lot of extension into that space in order to bring more access to people who maybe do not necessarily have it, who live in areas that health care entities are like far away from where they live.

You know, you and I had talked earlier and one of the things that left me a huge impression about you was that, you know, you like leaving things better than before. And so, you you actually made the statement, I like problems where when we're done, the world is a better place. And so, where do you think that next better place is from your perspective? Yeah, I you know, one of my big passions is trying to make sure that everyone realizes that for every friction point that we think we can design away with AI and technology that isn't designed away, it's going to fall onto a person to absorb. And, you know, that's always been a very key driver for me on how we're going to make health care better is understanding those kinds of friction points for people that who I when I went into health care and I graduated, 99.9% of people who leave health care schools are passionate about helping sick people. Like they want to get in front of people who are not feeling well to

Make them better. And that's where they get their reward and their fulfillment. And then there are all these people who are quite sick and would really like some health care to make them feel better. The idea that we can design useful and effective tools to help match those people together and that they can just focus on providing health care and receiving health care and everything else can kind of feed into the background. That's the kind of health care system that I'm I'm excited to build and I think that's how it'll become better is when people can just focus on either helping people feel better or feeling better themselves, and everything else can kind of be absorbed by technology. Or even both maybe, you know, at the end of the day, too. So.

28:59 Yeah, which is great. And I think that's a great close on the content part of our conversation today. I do have one final question that I ask all my my guests, and it revolves around, you know, what you've learned, what you've experienced. And so, if I were if if current-day Vishal were to go back to 22-year-old Vishal, and you could only provide one piece of advice, whether it be personal, professional, financial, whatever it might be, what would that one piece of advice he would give your younger self knowing what you know today?

I think what I would tell myself is you have a lot more time than you think you do to keep learning. You know, now with with growing your your family, free time is a lot more at a premium, but I think that there was a lot more I wish I would have learned back then and experimented with. You know, I dabbled in things like software engineering and learning how to develop my actual coding abilities and getting my hands a little bit deeper into the technology side itself. And I wish that I had, I think, spent a little bit more time on the clinical side, as well, more direct patient care.

You know, I from what I can see uh with my peers around me, there are a lot of technologists that enter the healthcare space. There are a lot fewer healthcare providers who enter the technology space. And what I think that helps bring is that voice of the patient into our development. And I think that that's like invaluable to have. You know, a lot of the the stories and anecdotes from the experiences I did have clinically still inform a lot of the design that I do. And I feel like the the experiences maybe that I could have built, the deeper and more robust my my solutions might actually still be today. So, I think that's what I would have told myself. Like, keep practicing things a little bit. Like, get your hands a little bit more deep into the into the software side of things and uh you know, just keep learning. Just just don't stop, even if you're tired, cuz you you don't know what tired is yet.

Well, I have to say what a great perspective and I love what you just said around taking that clinical experience and bring it bringing it to the technology side where you're designing the patient experience or the provider experience or the clinician experience into the technology, which for the most part for years and years and years and years was the biggest challenge on having wonderful functional technologies, but not having the right UI or UX that actually created, you know, a frictionness environment for people to adopt those things. And so, you know, what a great perspective. And so, Vishal, I really enjoyed having you today. I I'd love to have you come back and talk about more technology in the future a little bit as well. So, I'll leave that door open. Thank you so much and and great seeing you.

31:51 I appreciate you. Thank you for having me on. I really had fun. Thanks. So, let's connect. Let's find where your revenue is getting stuck. Book a call at RevRx AI and build your revenue engine.