This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

Newsday: Epic’s Growing Ecosystem and Today’s AI Realities with Srini Koushik

This episode is brought to you by Rackspace. The cloud is revolutionizing the way that healthcare organizations operate while delivering improved patient care, increased efficiency, and cost reductions. Cloud based healthcare solutions from Rackspace technology makes it easy for your organization to access patient and research data, collaborate with others on the continuum of care, and scale up or down as needed.

Rackspace Healthcare Cloud allows your organization to securely store, process, share, and analyze clinical information. Let their experts help you determine which private cloud or public cloud is ideal for your workloads. Check them out at ThisWeekHealth. com slash Rackspace.

Bill Russell: Today on Newsday.

Srini Koushik: At the end. really not about all of the technology and the mechanics of moving patient records around.

It is about quality of care. It is about access to care. And if the technology and what they do enables that, absolutely, why not, right? Why wouldn't we go out and do that? My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of This Week Health. where we are dedicated to transforming healthcare, one connection at a time. Newstay discusses the breaking news in healthcare with industry experts

Bill Russell: Now, let's jump right in.

right, it's Newsday and today we're joined by Srini Koushik, who is the President of AI Technology and Sustainability at Rackspace Technology. Shrini, thank you for taking the time. This is actually a global conversation because you're actually in India right now.

Srini Koushik: Yeah, I am. I, got here a couple of days ago and working on some of the cool initiatives that we've got going on here in India.

So it's it's a pleasure to join Newsday.

Bill Russell: forward to the conversation. So you also have sat in the CIO chair Magellan Health. If people aren't familiar with behavioral health, I'm familiar because I lived in St. Louis and Magellan was a. Pretty large employer in St.

Louis. Yeah. I'm curious, AI technology and sustainability. It's pretty pretty broad. What do AI initiatives look like at Rackspace out of curiosity?

Srini Koushik: Great question. AI initiatives at Rackspace covers the spectrum, but if I had to narrow it down, it came to two different things, right?

To explain that if I take a step back, what is Rackspace known for, right? We're the company that you work with in what we call day two and beyond, right? Once you've built an application and put it into production, that's when the real life of an application starts. We're the guys who are there throughout the lifetime of the application.

And so the core strength of Rackspace is knowing how these things work in production. And what we're starting to do is to figure out, okay, if you know that you can build applications that are designed to run and then that's, in a nutshell what our strength is. So in AI, especially given the newness of it, a lot of people are getting into AI and they don't really understand what it takes to operate it.

They get excited about building the use case, but operating it becomes a lot more difficult, keeping the the solution so that you don't get model drift and you start losing the fidelity of the model. Those things are all major issues that you have to deal with once you start putting it in production.

So we decided to take our learnings, Take what's going on in AI and say let's build these AI solutions that are designed to be industrialized and they can operate. And so that's the core of what we do in AI. So we do go out and build. new use cases in industries, including healthcare. But once we built it, we actually do work with the teams to help you operate it in the environment driving forward.

It's interesting when you

Bill Russell: say, what is Rackspace known for? I would say back in the day, you were known for data centers. Like you were a hosting, then it moved to cloud. It's like, all right, now you're a cloud provider. Now it's moved to more workloads. You're one of the largest.

Epic workloads, like you're handling more epic workloads than anybody else the world.

Srini Koushik: And that's absolutely true. But while that may seem like a switch in directions, it's really it's the right evolution, right? Because. Data centers, then the internet takes off and what you had to do was to simplify the process of managing that, that is managed hosting.

Again Rackspaces is there to do that piece. You start following the evolution over the last 15 20 years, then it was cloud and how do I actually start providing those clouds? The common thread through all of these things is workloads. Because when you design a workload, whether it's an application a data workload the gap that exists in the industry is people know how to build applications or data workloads.

People know how to support infrastructure. There's very few people who know how these applications are supposed to run on the infrastructure, or how do I fine tune the infrastructure to support this particular application? And the 25 plus years of experience we've had, that's where our superpowers are.

I know we said this is a niche in the market. This is a very much needed area of support in the market. So that evolution of workload. So you will hear us talk about workload aware modernization. You don't, Pick the destination and say this is public cloud or private cloud and then decide what to do.

You understand the workload and decide what the destination is needs to be.

Bill Russell: As we look at stories we'll talk about Epic a little bit, but I want to start with this one. It's it's a post from Allie K. Miller. So Allie, she goes by AI Allie. She's one of the most followed. People in on social media with regard to AI.

All right. So it's gone up significantly, but media and entertainment, it's it's listed as a risk, 92 percent listed as a risk. Software and technology, 86%. Telco, 70%. Healthcare, 65. 1 percent are saying. Look, this is going to change the game. We believe that a risk to our business.

Going to mention it in our annual report. that consistent with conversations you're hearing and you anticipate?

Srini Koushik: Absolutely, right? you think about I know it's an overused sentence, but it's with great power comes great responsibility, right?

The Spider Man quote that everybody uses. AI is no different. It is like in what generative AI and what large language models and what they've enabled is they've opened up tremendous possibilities that we're actually looking at. A wide spectrum of really things that we thought never possible before that you can actually do, right?

That's the great part, right? However I think in the rush to be able to go use that, there are several aspects of enterprise risk you've got to consider. And if you're so the great part. is where in healthcare we've always tried to figure out how do you improve access to care.

We've always tried to figure out how you can deliver quality of care because most providers, most doctors are doing a great job. They don't fail because they don't know what they're doing. Science is moving so quickly that it's tough to stay abreast with it, right? If you're thinking of quality care and other types of things, it is just that ability to keep track of it.

AI has the power to be able to help you do those things and bridge those gaps. But at the same time, the risk of AI comes in, there's a level of core competencies that's developed in everyone, I read a different article where it said, The Gen Z folks are absolutely phenomenal at using technology, but they don't know how to type.

And I thought about it, I said, that's a funny thing. Let me go read that and see, and it's actually true. What reminded me of is you are so good at when you had to type and communicate, like the haptic response that you had with typing meant allowed you to learn and others, and as these smartphones have come out and the next generation has come out, I start to see some of these text messages and I can't read it because they're simply emojis.

This, it's a different language that's evolved. Not a problem. The issue is if most of the corpus knowledge is built on the traditional model, you're going to lose the skill to read. They make it akin to this generation there's not that many that can actually read an atlas.

they don't even know what a cryptic is those are the things that you and I are probably used to.

Bill Russell: What is an atlas? Is an atlas, is that like a whole bunch of maps put together? What's a map?

Srini Koushik: We joke about it, , just 20 years ago in the 90s, we were using atlases.

I'd pull out that stuff to be able to go out and navigate from one city to the other. But in 20 years, that's gone. The generation that was born after that, they've never seen one of those things, and unless they took an interest in it, they never read this. The downside of AI, and that's why it gets back to enterprise risk, is as you start relying on these things, you lose that fundamental ability to know what is Because you're, you've shifted over to, believing everything that the AI is telling you.

So that has potential of risk from wrong answers to job displays and everything else that's there. So it's not a surprise that it shows up in every annual report these days. It's something that most boards are wrestling with. And most executives are optimistic, but they are cautious right now as to let's figure this thing out.

Bill Russell: So if I'm a healthcare CEO, I'm listing this on my annual report, and my thought process is probably if I thought about it, more the business side of it than the clinical side of it at this point. And the business side of it is, look, if our competitors are able to establish a different cost basis than us.

If they're able to put cameras in every one of the. patient rooms, and they're able to go from, a nurse ratio of one to five, and all of a sudden now they're at ten to one, and they're able to reduce their claim denials by five percent, which is millions, it's tons of money. They're able to do that because AI is actually reviewing all the claims before they get submitted.

And potentially AI is processing those claims and checking the medical record and all those kinds of things. If they're able to do summaries for critical patients that used to take. of research of a degreed person and the systems able to pull all those things together.

All the different modalities of AI being used we're not even talking about imaging yet and whatnot. We can create these new baselines of ratios and more highly efficient schedules. And I'm able to see more patients because ambient listening is there and doing its thing. And that's the reason I'm listing it as a CEO.

Cause I'm sitting there going, man, if the health system across way, all of a sudden figures out a different cost model and they're operating at 7 percent margins and I'm operating at 1 and 2%, They're going to be able to invest in that new building, they're going to be able to invest in more, key investments in physician practices and other things, and That's going to put me at risk.

Srini Koushik: Those are absolutely great points and my thing is the way I describe it is those healthcare providers, payers who I can see, can figure out how to use AI to deliver exponential results because the things that you're talking about and the use cases that you've listed, you listed several really good ones.

There's always a shortage of talent can we actually leverage AI to pick up some of the tasks so that the nurse practitioners and whatever role you're they can actually focus on more important things so that AI takes care of the toil. That's one way to go out and do that.

You talked about reducing claim, the claims denial and others, many times, like some of the coding errors that happens when you actually do drive that drives the back and forth, like you have the ability to put AI in line to fix it and drive that.

So the potential to drive that is extremely important. I think you brought up a really good point because the. RISC that I was talking about versus like the opportunities. I'd say there's one thing that brings that risk in and that's the generative component of AI like when you're doing transcription, ambient listening to be able to transcribe that's fantastic.

That's not the higher level of risk, but if you're actually starting to write care plans based on things and using generative AI to do it. These models are getting very accurate, but a 0. 2, 0. 4, 0. 5 percent error rate is not acceptable in healthcare.

Bill Russell: on the clinical side, for sure.

Srini Koushik: And so that, that generative AI component of it, like when I see some of the Fanciful here's the future, we can do this, and we can do that, and it's yes, you can do it, but should you do it? Because the mistake of a type 2 error or the mistake of doing something, calling something wrong, when it does come to the healthcare of an individual, it is far more consequential, the risk.

for a healthcare provider is much higher. So that's why this is not a cautionary tale of don't use it, but I do think that there are several options. I know a few use cases where AI is being used to simplify in one of the industries that is at risk of doing this is revenue cycle management.

You already know that there's these large companies that deal with just Let's code it, let's do it, let's do things so that the denials between providers and payers goes down and like we optimize that every cycle management. Something that AI is very good at doing. It can learn and drive that piece to be able to build it.

So there are so many areas in the healthcare value chain. That is today fragmented where AI is going to transform that. And as a CEO, you're absolutely right. I see potentially disruptive things coming in where people figure out how to use it better. The risk from existential risk and all of that stuff actually comes from more the careless use of generative AI a more yeah.

Use of generative AI comes in there.

Bill Russell: You're in India. You're working with some development teams around this AI one. we're going to get to Epic in one second, but I have to throw this one out there because it feels like marketing to me, Amazon CEO, Andy Jassy says, companies AI assisted has saved 260 million.

and four and a half thousand developer years of work. It's been a game changer for us. That seems a little marketing esque. Do you really think it saved them 260 million and four and a half thousand developer years of work?

Srini Koushik: See, I think Any organization has got that piece coming in and I'm sure they've got data that backs that up.

Here's what I would say about that topic though, right? I think if you've been a CIO running things, you know that you have what we call 10x developers. These are the smartest developers we've got. These are the people who can actually produce amazing results. And if I'm a CIO, maybe I have 5 percent of those 10x developers or even less than that.

The rest of them are really good people who are the 1x developers who are just trying to do a good job, but it's the complexity of coding and the challenges they have to face are there. What AI is helping us with today Is to get these one x people to act like five X or 10 x because the things that the 10 x guys actually already know here's the way to write secure code, right?

Here's the way to make sure that you don't create inadvertent inefficiencies in the code that you're writing. Something like Amazon Queue or GitHub copilot can actually help us with if used properly. So I think while people talk about it as. Oh, this is gonna change this.

It's saved a lot of money. I would say the savings for us and the customers we're working with comes from the fact that we can use it as a paired programmer to raise all boats as we see it. That's where the efficiency is coming from. I'm sure some of that is factored into Andy's numbers, but again I think you take all of these numbers, you want to understand the context in which they're coming from.

Are you ready to get insider access to the latest health IT innovations? I'm Drex DeFord, and I want to personally invite you to one of our upcoming webinars, Fireside Chat, Cutting Edge Conversations with Top CMIOs, sponsored by Dr. First. This is your chance to hear directly from some of the brightest minds in health IT as they share groundbreaking insights.

from one of our 229 executive summits. We'll cover the transformative power of AI, strategies for optimizing healthcare operations, tackling physician burnout, and the latest in population health management. Don't miss this opportunity to Stay ahead of the curve and bring these cutting edge ideas back to your organization.

Register now at ThisWeekHealth. com slash Fireside Chat. That's Fireside dash chat. ThisWeekHealth. com slash Fireside dash chat. Thanks. See you there.

Bill Russell: That's how I talk to my team about it is this is your AI companion. So we pay for AI for every one of our staff members. This is your AI companion. And we're training them on how they can utilize it. I know for me it's 10Xing the stuff. Cause one of the things I do is I turn it into an antagonist.

Like I give it my ideas and I say I want you to ask me questions back about this, trying to debunk what I'm saying. And so we'll go back and forth. It's a really interesting antagonist to my thoughts and my ideas. And it helps me to fine tune them. I think a lot of people just naturally just ask it a question, take the answer and away they go.

But there's a lot of different ways to use it, and it does

Srini Koushik: And what you talked about is actually a really good way, right? Because you get into this confirmation bias, and whatever you wrote, you start to believe it. But having that ask it to play the role of an antagonist who doesn't believe in what you're selling and poke holes in what you've written up like that.

It really does go in and so you start seeing some of the downsides that come in. That's the example uses, that's the perfect 10x thing. Because are other places where tell you my version of that is I do what you suggest, but my first language is not English, right? However, I've got ideas that are there and like most of the times prior to having generative AI I spent time just forming the sentences, getting it right, all of those other things.

Nowadays, like if I have to write something, I go write my four points. These are the six things I want to be able to write. And I then go tell it to write my first version. And the first draft that it produces is way better than I would have done, by doing it by myself. That's it, right? I take the first draft and then add my version to it.

That's my version of 10 x. Because if something, because my language, first language was in English if it took me 10 hours to write it, now I can finish that whole thing in an hour. Yeah, that's a 10 x improvement for me. But that's the beauty of the AI to me. And that's the potential where every human being is different.

But if you can train them to figure out how to work with AI, now you have a lot of people who actually do what they do, but do it much better, much more efficiently.

Bill Russell: All right. Let's talk about Epic a little bit.

I don't suppose you got a chance to go to UGM because you're overseeing it.

Srini Koushik: I've done it in the past these days and I don't get to, but it's Epic's a great company. They built something pretty amazing and it's always fun to go to those conferences,

Bill Russell: Yeah, so AI was everywhere, right? The cool thing is how they're listening to clinicians and their clients, and they're prioritizing different things.

So they have this idea of like patients or similar patients, and they're able to do diagnosis based on that. This is something that clinicians have been asking, for years. They're using Cosmos to find the Patients across the country that have very unique diagnosis and those kind of things.

It's patients who are that 1%, but there happens to be somebody else, and they're able to find that. That was interesting. Obviously, ambient listening was everywhere. It's not only nuance anymore. It's nuance, it's a bridge, it's ambience, it's da. There, there's a bunch of them playing in that space and some of them have deep integration to Epic these days.

Chart summarization was another thing they talked about. Which is a real game changer being able to really, to go across that longitudinal patient record and bring all that stuff in. Bye bye! That is relevant.

Srini Koushik: Cosmos to me was fascinating and actually something that's a lot more practical because it does play to the strengths of what AI can do, right?

Look at that entire spectrum of things and if I can, as a doctor know, I've made you a the doctor, a 10x doctor because they're good at what they do and they're But they don't know what they don't know yet, right? But that problem probably has been solved and treated in a different setting in this whole thing.

Having Cosmos I love the way they call it. It's like patient. It starts to give based on certain hypotheses, starts to give some recommendations, which then gets the physicians think, I didn't think about it that way. Let's go out and drive that. Cosmos, I thought the way they described it was ideal and I can see how physicians would absolutely benefit from a tool like that.

The ambient listening was interesting because the ambient listening and others, I think about it as, There's that whole shift going on from typing to typing in mouse and all of that stuff. The whole modality of interacting with computers, especially now that they can understand natural language very easily, the modality of how we use it is going to change.

And the ambient computing and others, like the, now the technology has caught up so that it actually can be very accurate and it can save a ton of time in doctor's offices. Because every time I go to a doctor's office, the doctor will spend time with me, then he turns around and types something into Epic and then doesn't speak to the whole ambient.

Srini Koushik: listening for the physician so you can pay attention to the patient, but also the back office functions that have to be done after the visit is done saves a tremendous amount of time and allows them to focus on the patient. So I can see how that is relevant.

think one of the most interesting things is that they're expanding, right?

So they now have a payer platform. They're doing blood bank. They're doing a thing called like Compass Rose, Yeah, Compass Rose, and really that whole population health thing. They're connecting up these agencies and food banks and stuff so that they can participate. And I think the other thing from an expansion standpoint, big international push, like they are pushing internationally.

They're getting to be more of an ecosystem that could really impact That whole flow from insurance carrier to provider to ordering tests to, they're really starting to encompass that whole thing.

to some players, it's a little scary, but to others, you're like, man, the opportunities there to create efficiencies are. Really great.

Srini Koushik: They are and you stated it very accurately, for people who are not in healthcare, the U. S. the healthcare system is a very good one, but it's also a very fragmented value chain with so much inefficiencies in this whole thing in terms of this whole payer provider, PBM, all of this interaction and all the middlemen that have to make this work.

system that works, but it's very inefficient., that prevents us from doing what we do. I can see how Epic who started in the center saying look, it's all about electronic health records. Can start to say I can start to solve some of these inefficiencies in the healthcare, if you get that, and there's a lot of benefit.

Srini Koushik: And this is now we're back to the all over, is the Apple wall garden better, or is the Android open model better? That's the way I look at it. I said There's people like me who lived the Apple ecosystem that won't get out of it because I don't mind the walled garden, because I get things that work seamlessly and track all of those things.

And then there's others who will swear by the other one that just says, I just like the flexibility. We're going to get into that, but I can see why Epic does it. And I think it's a good move on their part to solve it. At the end. It's really not about all of the technology and the mechanics of moving patient records around.

It is about quality of care. It is about access to care. And if the technology and what they do enables that, absolutely, why not, right? Why wouldn't we go out and do that?

Srini Koushik: It's nine o'clock. At night? Nine o'clock at night.

Yeah.

Bill Russell: Thank you. Thank you for spending some time with me tonight and appreciate it and look forward to catching up with you soon.

Srini Koushik: Thank you so much for the thing. And then for people like, Epic is one of our biggest workloads from a Rackspace standpoint. We are the second largest hoster of Epic workloads after Epic themselves.

So we're huge believer in this thing. We're actually committed to. To make sure that vision that Epic's talking about, we're part of that, part of enabling that. Appreciate you taking the time with us and appreciate you giving me some time to talk about my thoughts on AI.

Bill Russell: Absolutely.

Absolutely. Thank you.

Thanks for listening to Newstay. There's a lot happening in our industry and while Newstay covers interesting stuff, another way to stay informed is by subscribing to our daily insights email, which delivers Expertly curated health IT news straight to your inbox. Sign up at thisweekealth. com slash news.

Thanks for listening. That's all for now