Hello everyone and welcome to ES Speaks, the show where we unpack the biggest ideas shaping enterprise strategy and performance. I'm your host Corey Nolles and today we're exploring the AI prescription for a better patient experience. Healthcare organizations everywhere are under pressure to deliver faster, more personalized, and more connected care all while maintaining trust, compliance, and security. But how can AI actually help them get there?
To help us answer that question today, we're joined by John Matthews, industry strategist for healthcare and life sciences at Terod Data. John works with some of the world's largest health systems and payers to design analytic strategies that improve outcomes, reduce friction, and scale innovation. John, welcome to ES Speaks. >> Thanks very much for having me, Corey. I look forward to the conversation. >> Good. Me too. Me too. So John, I I guess let's start with kind of the big shift.
Why why is patient experience emerging as the next major competitive differentiator in healthcare? >> Yeah, you know, it's an interesting question for for those of us who've been in healthcare for a while. I think that um it's always been um really important but I think the the emergence of it um in this in this framework in the ecosystem in which we talk about it more is because we have AI we [clears throat] have some technological capability uh that we didn't have a number of years ago and so there's great opportunity to improve the space um it's critically important it can mean the difference between keeping a patient losing a patient on the competitive side um but I want to say like in The biggest picture of all patients have to participate in their own care.
If we're going to maximize outcomes, if we're going to limit cost, right, health is determined by what happens outside the four walls of the hospital and then what happens when you're engaging with the hospital or the doctor. And so awareness um being informed and knowing how to navigate are critically important and we serve that through patient experience. >> That's awesome. And yeah, it's definitely where it has to begin. when we when we talk about like the AI powered patient experience, what is what does that really mean?
Where are we seeing the most impact today? There's there's a couple of areas that I think directly and indirectly impa impact patient experience, excuse me. I think uh when you see um patient navigation inside of a hospital system, I I think it's a large gap. Um, but we're starting to we're starting to close it with better um more prescriptive direction and and and we're having a better interaction with the patient when we use um smart chat bots when we can route them faster to the right people uh when they have a question.
You know, I like to say um healthcare is like no other industry. People don't pick up the phone and call their physician or their insurer because everything's okay and they just want to engage with healthcare, right? there's a particular need. Um, we see things in in the hospital system and like digital front doors where we actually have like a more traditional sort of retail approach where you're trying to you're trying to take a doctor shopper and convert them to an actual patient.
That means patient revenue, but it also means that that patient who has a need is getting connected with somebody who can help them. And so I think areas like like navigation and digital front door, those are areas that we're going to see um direct involvement with the patient. But there's also all of the back office, right? Traditionally, health care systems are systems and smart software developers and people patching things together to make it work well.
Now with with AI and and the advent of language models, generative AI, there's a lot we can do to make that better and faster and in some cases cheaper. I that's awesome. I agree. I agree. And and how do those like patient expectations, what they're expecting when they come compared to what most health care systems are currently able to deliver? >> That's an enormous gap, Corey. It's just, you know, it's just an enormous gap. Um I think a lot of us don't understand how complex and sophisticated healthcare is to begin with.
And then on top of that, you you just there's a there's just an enormous lexicon and awareness gap. It's it's an asymmetric environment. Most of us don't know how healthcare works as a system. Most of us don't know clinical terminology. We don't know what the doctor's saying. If you've ever taken a look at a at a pathology report because you had some sort of biopsy, it does you don't have you have no idea what it says. And so this gap is just enormous.
And there are ways I think to to level it out a little bit and I think that's where the greatest opportunity exists. Um but we have a [clears throat] long way to go. >> So what are the what are the like biggest risks for organizations that are falling behind and fail to modernize in this area? >> So the business risks are competitive right lost patients lost revenue. um you know even on the insurance side right we in this country we we typically tie our insurance carrier to our employer and our benefits manager and the choices that happen there the but in a world where we are trying to attract the best talent the benefits matter and so employers are listening to their employees about how they're treated what their experiences with insurers um so the so the business risk can be can be lost members it can be lost employer groups for payers it can be lost patient patients uh on on the the hospital side.
But in in the grand scheme of things, the inability to modernize in this area is going to lead to high cost and worse outcomes. We're on a trajectory to just continuously spend more and have a less healthy population. That's got to change. >> Yeah, 100%. And uh I I think this is an interesting approach. Well, let's uh let's talk a little bit about the barriers to reaching that better standard of care. U many health care organizations really struggle with fragmented EHR systems, data silos.
Why does this continue to be such a persistent problem? Well, from a technology perspective, the way we designed systems in the past very much locked the data with the EHR system or the proprietary system in in the point solution. So, you had the data for the EHR. The EHR contained the patient record, but a patient record is just data about a patient. Why does it have to stay and live there? Meanwhile, we've got vendors who have lived a world of privacy and proprietary um to control that space.
Some of it appropriately and some of it a natural sort of business defense if you will and they don't really want to play ball necessarily with other vendors, their security there. And then on top of it, you've got a market where people leapfrog each other, others fall away and so you've got a smattering of systems and that leads to that kind of fragmentation. The other thing is it's really expensive to then get the data out or harmonized across all these different systems.
It's just a very large um an operating health care system watching data flows is uh is is like a hairball and to deconstruct that hairball and to make it make sense and to turn it into something that is modern and enabling modernity is uh is quite expensive. And so I think uh that's the primary challenge that we see. >> That makes a lot of sense. So how how do these these data challenges for example directly affect patient outcomes and operational efficiency? >> To me fragmentation of the data is time loss at a minimum.
Right? I mean to to make one system give data, another system give data, tie it together, harmonize it, make it make sense, and then get it to a decision maker is time. It's time, it's energy, it's, you know, it it all has to be quality checked. Um, that's lost time. And and when you're racing against time in some cases because either it's clinically warranted or you just have a patient in need and the patient's asking sometimes you're going to make decisions best guess.
Now that becomes um that's not the best situation for the best outcome and it's not the best situation for the patient. So um so not having these systems in place and not having um a fabric that is easily and readily accessible and enabling to to modernization it is really a time and efficiency loss. >> Wow. So what role does regulatory uncertainty especially around AI play in maybe slowing down transformation? So, I don't think it's ever a good thing when regulation trails technology so far.
There's always going to be a a time gap, but when it when it's this far behind, um it becomes ambiguous to a point of a real hindrance. U imagine a lot of these systems are setting up governance bites where they're trying to figure out what does AI look like for us? How do we really use it? Where does it fit? Where is it dangerous? How do we govern it? And then when you add into those conversations a risk officer, um, corporate counsel, outside expertise, you start to have personalities that can dominate the conversation based on that ambiguity.
They are people who are more riskaverse than others. By not setting the standard, we don't know what good looks like. We don't know what the target is. And so it really does you lose time in debate, but then you might lose the opportunity in the decision because you just didn't know. And because you didn't know, you took the the most risk proof way um path you could, which is just don't do it. We'll wait for we'll wait for direction from the authorities.
And that really slows you down. >> Yeah. Yeah, I'll bet it does. Especially in a field like healthcare where you have so many requirements around explanability and all of that. So how can healthcare leaders balance compliance requirements with the need for speed and to move forward with innovation? >> I think there are a couple approaches. Um first is you absolutely need the governance body but I I think you need to practice the discipline outside the space of risk.
So things like digital twins, development environments, simulation environments, those are the places where you can really practice, let's call it the um the discipline of the technology and putting AI in place and seeing what it looks like and talking with your physicians and your clinicians and your administrators, you can start to figure it out while building the muscle groups that you need to do it in real life in production. And so some areas I think are going to be relegated to that simulation environment to make sure that you're able to move forward.
There are other areas where it's really just about process. You know, healthcare likes to think it's got a lot of proprietary process and every hospital is special and different from something else. It's not really the way it works. Yeah, there's some there's some particularities, but generally speaking, patient care is about a patient seeing a doctor. There are notes. There's a medical record. There's a billing process. All of this is is open to reinterpretation with AI to make it faster.
Our goal should be to have our administrators, our caregivers working at the highest level of their capability or the highest level of their degree. And to do that we need to offload um the burdensome side, the manual data wrangling side um you know insert name of process that that really can be agentic um here and we'll go faster and and we'll build trust that way as well. >> That makes sense. and and you know it's definitely the smart approach to hit the things that are more adjacent to patient care than than actual issues where that explanability comes into play so much.
Let's uh let's talk a little bit about infrastructure. You've you mentioned the importance of governed and scalable AI infrastructure. What is what does that look like in a healthcare context? Yeah. So to me, and it's probably not unique to a healthcare context, is the the data is prime. Um the the way the world is going, it it's an information age. It is now an AI age. And all of that requires data. And so we've got to get away from um a perspective of of coded application to act and sort of think about when that happens, data is created.
How do we govern that data? How do we harmonize that data with all the other data? And where does it sit? Right? So the separation of data from um its specific purpose by actor or system or user I think is really important, right? And it's this it carries through a consistent theme with with like cloud separating compute uh from storage, right? let's separate data from data operations to make it available for any operation. And I think I think that's the first key in setting this up.
And then the rest is I think more traditional technology uh computing and infrastructure and and systems and people and discipline and governance. Uh but but I think it's the data first approach uh that we really have to use in this era. That makes that sounds sensible and and it's very much a common theme we hear in other industries as well is that you got to get your data house in order. Uh got >> or or you're going to have real troubles. How does um how does a cloud-based analytics architecture help unify systems without forcing a provider to rip and replace all of their existing tech?
Yeah, I I tend to I think of cloud in um in a in a modular sense, right? Because we've because we have separated the decision making of what application or thing we need to do from whether we need to order a server and how much memory it needs to have on it and how we rack it. You know, the moving the data center was was a big thing. But if you if you follow that line of thinking, what we have right now really is kind of a plug-and-play environment.
And and that's what healthcare needs, right? We're in a period of rapid innovation. And so in a cloud space where they can quickly spin up whatever infrastructure you may need and computing power for the intellectual property for the third party actor for even their you know their native new release of XYZ function that that is so much faster um and so much friendlier to more antiquated environments that you know we can't go do a big bang conversion on we don't have the time we don't have the money and and you know the old the old joke about uh you know war fighting in and the Pentagon in the United States is we're always training to fight the last war right when you do a big bang approach you take two three years to do a major migration and conversion and three years later you're ready to go on the problem from three years ago >> awesome well done right >> it's got to be it's got to be incremental it's got to be >> because you have to course correct along the way Yeah, that makes sense.
It's very much a um [snorts] it needs to be an ongoing all the time thing, not a uh we're going to do this and then we'll start X when we're finished >> and and where do we got to go next? Because something took a higher priority today or the market has shifted or are we had an epidemic and a pandemic and what did we need to do then? So I I think I think cloud allows for that flexibility and and those course corrections and that incrementalism uh that I think is so important to progress. >> Yeah, I agree.
Um would you mind sharing how terod data and AWS are helping healthcare organizations to build like flexible compliant data ecosystems that actually work in practice? >> Yeah, sure. Um I I think um you know I've spent some time talking about data is prime and when I think about terod data that's what I think about I think about our mastery of of data and all the different ways in which you you need to do something with it to make it available to um to exploit it appropriately for the benefit of of your customers or or um you know your patients or your members how however you wanted to.
Um and and when you introduce that sort of specialty with cloud and and cloud to me is a is a euphemism for execution as we've just talked about speed of execution. And so when I can take um the best of of what we have to offer in our in our deep industry knowledge of of data and and data modeling um our our foundational tools around analytic models and model management, our ability to um to take third-party analytic IP and and push it down into our computational massively parallelized system for speed.
You know, that is what Terodate is bringing to the party. And when you think about all the things that AWS brings to the party, I mean NLP, Comprehend Medical, Bedrock, Sage Maker, right? All of these points of um of integration that that Terod data has worked hard to make available with AWS. I I think you've got a very very powerful prescription and we see that in the number of customers uh that we work with. Um and I think if you if you ask them about um the speed with which they can address challenges once they've done that um they'll they'll tell you it is it is much better for them.
It is much more flexible for them. Uh and and when they fail they fail quickly and can move on to the next thing and rip and replace a small piece of the puzzle instead of the entire environment. And those are good things. >> Definitely. Definitely. Do you have any real world examples of how this approach is improving patient experience whether that's through predictive care, faster diagnostics, even better engagement maybe? >> Yeah. So, I I think um I think the the the the most direct I I alluded to this earlier with like sort of digital front door, but being able to convert um being able to convert doctor shoppers and and and patients and looking and seeing kind of what they're looking at is a diagnostic tool in and of itself. and the speed with which you can get that to a um a disease manager or a navigation expert um to get you to the right point or an appointing person who who can think about priority is um you know is is pretty critical and we've seen that we've seen that in places where we have an unknown patient and we can figure out how to get them in the door um quickly and we've seen that with a known patient who's searching up oncologists.
I want to know I want to know if somebody in my care is searching up, you know, cancer care. >> Yeah, >> that's a that's a pretty important tell. Um, that's that's worth an outreach. >> So, that's where we see the power of of AI and analytics and and and chat bots and the information being available to just speed the process along and get people to the care that they need to get to. And that doesn't even touch on on you know other areas of the of the ecosystem where we see what image processing can do um in in X-rays or lung nodules to to more quickly identify real clinical problems.
Um, you know, that's an area where, you know, human in the loop always, but the the pixel color gradation difference that's mathematically recorded in an image is much easier for a computer to detect >> than it is for the human eye. Right? So, we're going to see a lot more of that more discreet stuff happening. >> Um, and we're going to see it in imaging. I think we're going to see it in um in pathology, in slides. uh we're going to see it in in we're going to continue to see it in drug development.
Like there's a there's a lot of stuff where um where either we are um are directly impacting things or even if you just take um you know terod data as a facilitator of research recall. >> Yeah, >> there's a lot of data just around research. Call it you know the metadata. What were the parameters? What was the study? What happened? Well, it was for purpose A. But now we've got it inside of a a system with a with maybe an unsupervised learning algorithm or or an agent who's looking for other opportunities to reuse that information or that compound or that study for another purpose.
We're just exposing a whole lot more capability more quickly. Uh in that respect, >> absolutely. Well, let's on that note, let's let's dive into kind of the path forward a little bit and look and look ahead. Where [snorts] where do you see the most exciting opportunities for AI to elevate patient care beyond here? >> So, I think uh you know the the realist in me says right now it's going to be process and that's appropriate. That is absolutely appropriate. the this is still I mean in the grand scheme of the world this is still relatively new um but but process and reshaping I think how the system um learns to leverage AI is going to be the most critical and the most impactful over the next 3 to 5 years right we're going to we're going to go through some ups and downs but but that's that's where we're going to see um major shifts in just the operational efficiency >> and I think for and I think for a number of physicians if we do it right we'll see less burnout >> that's amazing >> we'll see less administrative burden on them I mean we already see it with with ambient listening agents and and scribes for for patient engagement feeding you know AI driven summaries and synopsis afterward we're going to see I think we're going to see a lot of process improvement that just makes things faster how many times have you been to the physician's office and you have to fill out the exact same piece of information >> and you think you know my history know this by now and part of that is part of that is is a safety protocol because of because of a lack of trust potentially in systems or because a mistake has happened or we've had an adverse event but I think we're going to see a number of those things practically improve um you know the the hopeful person the dreamer in me wants to see uh wants to see AI really change that that information asymmetry.
I want I want patients members. I want them to be a little bit closer to understanding how the system works and the system a little closer to patients so that it really thinks of of humans as an end of one. It's it's deeply personal thing healthcare. We need to treat it as such. We need all the statistics and the data on populations, but everybody is a person. And the closer we can get those two parties in terms of a shared knowledge, the much more efficient and much more effective our healthcare is going to be. >> I agree.
And uh and it's it's really an exciting field for AI to touch. What um what steps do you think healthcare organizations should take right now to prepare their data for scalable AI adoption? >> Oh man, the uh the maturity is so uh is so varied. I I think um yeah, this is going to sound so rudimentary, but there are a lot of systems that don't know where their data is. >> Yeah. It's just between mergers and acquisitions and just time and this system and that system and oh it's it's written on the mainframe or cobalt we don't have anybody to fix that let's just leave it it's working or there's just so much out there.
Um, I I think the the prescription I write is you got to know where all your data is first and and you got to have an adult conversation after you discover where it is about what what it what needs to change because you don't know. You're going to find out things you don't know and then stuff needs to change and that's an adult conversation about how you move the system forward. So inventory first and then again that design principle over the data is prime.
We we want to build a system that consumes the data we create for the benefit of our patients, our members and our and our and our employees. Um so I think it begins there and then obviously governance is um is critical part of the adult conversation is what what the governance committee decides. Um so I I think that's that's where I begin. Um is as as unsexy as it may be, >> yeah, >> that's where we are. We don't know where we don't know where all the bodies are buried and we got to go find them. >> That's and and it's a big problem across all industries right now.
Truth of the matter. It's it's very interesting. It's very much where you know it's the base of your Lego tower. >> Absolutely. >> If if you don't have that, you're going to have problems all the way up. >> Um >> once you have that in place, I think the rest becomes um you see the light at the end of the tunnel. your Amazon can help you build a data fabric. Terod data and Amazon can help you get all of that under control. We can help do all of the APIs into the various um AI systems or the algorithms or third party IP.
Like that's the fun part. It's the it's the plumbing and the infrastructure around that plumbing and the control mechanisms for it um first and then after that it's um you're into the fun part, you know, with with within within good guidance of of regulations and safety, right? How do you think health systems build can build internal trust and literacy around AI so clinicians feel more empowered and less replaced? >> Yeah, I think um I think it begins with leadership by example.
Um you know this is this is a very traditional approach. You've got to find your you got to find your stakeholders in your system who are um who are open. They are they are believers in potential or or correctly skeptical but evidence-based and can change their mind and so they'll participate in in trying this out. Um yeah, >> you know, I think about a I think about an academic medical center where I've got an attending physician on rounds with with nine students. >> Um and going, you know, patient to patient, bed to bed, having the discussion about the patient, you know, quizzing the students and whatnot.
I think of a 10th student, which is a which is a computer on a trolley. In this case, it could just be a could just be an iPad, right? Talking to the clava. But I I think all of that knowledge is available to AI. Why don't we ask the iPad and the 10th student to weigh in as well and see what they come up with. Now, is that going to be better than the other nine minds plus the attending? Probably not. But I would say I would argue that there are a lot of odd disease presentations and there's a lot of multi-year misdiagnosis, right?
It's not like a system rampid problem, but there's enough of it. and the knowledge base that AI accesses and how it reasons against all of that information offers, I think, alternative opinions that need to be considered. And if you've got an open-minded attending and you begin in that way with students, you start to build a culture that's more accepting of what it could do. Doesn't mean that it always will do something for you, but it could do those things.
And so, I think you got to find those open minds and I think you got to put it into situations where it could be potentially helpful. and diagnosing something rare and often misdiagnosed. That's an opportunity. >> It is. It is. And and I and I love the approach of of looking at it as an additional mind. I think that's that's very important. Uh and you know, the only winner there is the the well everybody's the winner. The patient, the health care system, the doctor.
I mean, you know, accuracy is always better. >> Yeah. for everyone involved on the health care scope I would say >> and just and the challenge of of thinking it through from another perspective right we all make each other better right it's diversity of opinion as well >> that's right that's right so if you had to summarize your AI prescription let's say for healthcare leaders uh what would the key ingredients be >> get started one you you just you have to you have to get started this is where we're going hop on the train get your data house in order and and be prepared for those for those critical, you know, adult decisions about what you're going to have to do and what you're going to have to remediate.
Um, and and and then uh, you know, get get your governance um, systems set up approp appropriately. Um, some of that is technology watching and is much more on the security side, but the the governance part of what you're going to do, how you're going to do it, where it's going to where it's going to make a difference, um, is really pretty critical and and you're going to have to do it from a business lens as well because some of these are pretty big checks and and nobody's just funded um, infinitely, right?
And so there are hard decisions to make about where where we go. And I think with those apparatus in place and and functioning um then then I think you're you're ready to go right and then you can you can call you know insert name of of of helping entity here whether it's terod data and AWS we obviously we'd love it for for it to be gospel but I think these are the things you have to do um to get going. >> That's how I would that's how I would look at it.
I agree and I think that's a that's a smart approach. [snorts] Well, that brings us to the end for today everyone. A huge thanks to John Matthews for joining us and breaking down how AI and modern data strategies are transforming patient experience from the inside out. John, before we wrap up, where can our listeners go to learn more, explore the use cases you mentioned? Uh you can certainly go to uh to ter's website where we've got uh numerous uh testimonials and and customer use cases in place and uh and I look forward to anybody who wants to reach out to uh to me or my colleagues who are widely available on the internet. >> Excellent.
We'll have a link to that in the description right below this video. If you enjoyed today's conversation, be sure to subscribe, like, and share this episode with a colleague working on healthcare transformation right now. Thanks for watching and we'll see you next time on ES Speaks.
This transcript was generated automatically from the
video's captions and may contain errors.
