Reimagining healthcare delivery by migrating critical workloads to AWS

Good afternoon and welcome to "Reimagining Healthcare Delivery by Moving by Migrating Critical Workloads to AWS." I'm Angela Shippe, Senior Physician Executive and Clinical Innovation Lead for Nonprofit Health Care. I'm an internal medicine physician by training, practiced as a hospitalist for many years, and also held Chief Medical and Quality Officer roles. So I have been an end user of healthcare technology my entire career.

I'm thrilled to be here today with Dan Bennett, the Chief Technology Officer of Geisinger Health, who's helping to lead their digital transformation.

Today, we wanna talk to you about:

• Who is AWS
• What we're hearing from healthcare customers
• AWS in healthcare
• And upskilling for the future

Now, I wanna get a sense for who's in the room with us.

How many of you are clinicians? Alright.

How many of you are chief technology officers, chief information officers, analytics? Alright. There's a few hands there.

How many of you are builders? You're at your organization's building and making this happen.

How many of you are from provider organizations? Alright, healthcare and life sciences. More health tech, fantastic.

Well, it's great to have all of you here and I know that all of us are consumers of health care in one way or another, whether that means that you directly have been a consumer of health care or indirectly with family or friends.

And so you might be familiar with the picture that's up on the screen right now. This is an operating room with a clinical team in it and you see a clipboard with a piece of paper. Now, it's been a while since there's been a lot of clipboards around, they've been replaced with workstations on wheels or they might be mounted to the wall. But the goal would be, can we get to this? Can we get to an operating suite where there is data and images and the real time monitoring of the patient right in front of them? So that really they're gonna just look up and in their line of sight is all of the information so that they can have an optimal outcome.

Well, at AWS, we're mission driven just like the organizations who are providing this care. For over 17 years, organizations have leveraged AWS's global cloud infrastructure with the broadest and deepest set of services to have a secure and truly reliable platform. And for over nine years, there's been a dedicated practice to healthcare and life sciences that has physicians like me and other specialties, nurses, researchers, pharmacists, chief information officers, analysts, all who have walked in the shoes of many of you in this room. And it is this group who's working in collaboration with customers, bringing the industry knowledge and technology together to be the trusted advisor to global health care organizations.

So when we have the opportunity to talk to customers, this is what we're hearing from them. Three big themes:

1. The decrease in margins. There's been a degradation. There are three major reasons why this is happening. One is the cost of labor supply cost and the other is the overall cost of care. Now, traditionally, these were easily managed. But more recently, there's much more volatility and variability and it's getting a little harder.

2. The workforce, workforce that's demanding to be in an environment and have the tools and training that's needed right at their fingertips, allowing them to work at the top of their license or the top of their skill set. And we know that when workers feel like they're doing their best work, that's a top factor in resilience.

3. And then lastly, what we've heard it is about the consumer, the patient across the continuum of care. Who's saying I now want an experience where I can achieve my best health, have access to care when and where I want it and be a member of the care team to ensure that that care is being delivered in a way that's meaningful to them.

At AWS Healthcare, we are invested to innovate on behalf of customers to really address those challenges I just mentioned and with that leading to the development of unique purpose-built services that are focused on what's important in healthcare and life sciences.

Take AWS Healthomics providing organizations with the opportunity to be able to take data, multimodal analytics from DNA to health records to medical imaging and to be able to utilize it to have more insights, improve clinical outcomes and really help advance scientific discovery.

How about AWS HealthLake a HIPAA eligible service that provides for secure transfer of transfer of unstructured data and unlock powerful query and search capabilities.

How about AWS Health Imaging, allowing providers and software providers to run their medical imaging applications in the cloud and with that, be able to decrease the cost and help improve the infrastructure.

And then lastly, how about AWS Health Scribe the opportunity for a vendor to be able to take clinician and patient interactions and be able to transcribe them and summarize those notes and make them readily available.

All of these services to make it easier to help achieve those outcomes that consumers want to achieve the outcomes that the workforce wants to help healthcare and life sciences continue on their mission and achieve their goals.

And we know that in the highly regulated health care industry, it's extremely important to be able to do this and a platform that's secure, that's reliable and helps to improve those digital relationships. And lastly can enable innovations like generative AI.

And we've been talking about that the entire week. So here's some key areas where gen AI can be used:

• For medical research, how about being able to fill in some missing data, automatically helping to improve those data sets, being able to enhance therapeutic development and identify rare diseases and accelerate the time to advance therapeutics.

• What about clinical efficiency? How about summarizing large patient volumes and increasing the time it takes to decrease the time it takes to do that, then you're allowing more access to care. Now, there's an opportunity to better understand the patient's clinical journey and personalize it for them.

• How about operational efficiency utilizing gen AI with operational efficiency provides that opportunity to take manual tasks, duplicative tasks and make them easier, automate them so that you can improve efficiency, decrease cost, improve the access. And this would be a great opportunity to bring in Amazon Q.

• And how about patient experience? How about using conversational chatbots agents to be able to make it easier for patients to get the information that they need and to be able to have it in a way that's personalized just for them. And then again, digital health where we're giving patients the access when they want it, how they want it, answering those questions any time of the day or night.

Well, in order to do all of that, there has to be an understanding of the cloud and not just the technical folks in an organization but everyone and there are lots of ways to get there.

So one is we have over 600 courses online and we're happy to help you curate that content for certain roles in the organization within and outside of it. For those who really want to accelerate their learning and receive certifications, we can help you with training with that as well. And then there's global conferences just like this one, like Re:Invent where you've had the opportunity to come in here in real time, what's next and to talk to each other and partners about how you can continue to evolve and improve your cloud infrastructure.

And just like it says here in this quote from Forbes, it really is about the people. It's more than services and servers. It's about upskilling the workforce to be able to sustain that ongoing development.

So everything I just talked about is absolutely possible. It's not just aspirational. And I would like you to hear from AWS customer, Dan Bennett, who is the Chief Technology Officer at Geisinger to walk you through their journey and what they're doing right now.

Please welcome Dan to the stage.

At this point, you would be hard pressed, walking through the hospital or walking through a facility and finding somebody that has not been involved in some aspect of this project, whether testing, providing feedback and other things there.

So one of the critical components in one of our additional transformation projects was around it's new operating model. So prior to this project and others, we were a traditional it shop. We had application accountability across multiple leaders. We had uh infrastructure teams, we had all those normal things and all their nice little silos and their stacks.

So one of the things that we spend a lot of time on is thinking about what do we need to look like for the future? What are the skills and how do we get our current staff? What they need to be successful? Because they were, they were, they were vested. They've been, they've been knocking things out of the park for years. And now, you know, we've come and say, hey, we're going to go to amazon.

So we had to really step back, do a skills gap assessment, find the right trainings. And again, those trainings were not just for the technologist, we put a lot of them through training, a lot of them got certifications, but we did executive cloud training. So people that want to know what is the cloud, what is amazon. So folks in the business were taking courses, things like that.

And then ultimately, as we did this and we're now in the far side where we're in this iterative build and constantly thinking about, do we have the right resources applied in the right spot at the right time. And how do we be nimble? So historically, when you think about an operating model, it's very rigid, it's very hr based, right? You report to who and so on and so forth.

We've gotten to the point now where we've kind of collapsed. Some of those silos, we have a single executive in charge of all applications. We have myself in charge of cloud and infrastructure and identity and access. We have all of those things now kind of spaced out where they need to be and we've started cross pollinating and collaborating across teams that normally would not have necessarily talked.

And we continue to look at as we go into an application reduction slide. Next, you'll see as you reduce your portfolio, you free up time for folks to start cross training and and cross pollinating those skills so that we can build a workforce for the future as we sustain the current.

So the second big piece and one that was tackled as we were in the design phase of the cloud. So this one kicked off very early for us in the overall project was around application rationalization. And to be honest, we talked about this for many years, we tried to do it many times. But ultimately, the way that the industry has changed and the headwinds and everything else, we were forced to get it done.

So we went through our application portfolio, we found duplicate applications, we found applications that weren't used and we went and we trimmed hundreds of applications out of our portfolio. We are continuing to do this today. This has now become an operational, an operational type activity. The business is involved or business stakeholders is involved, it is involved.

And as we bring on new applications that we need, we look to sunset other applications and as contracts renew, we look to make sure that there's still a valid application. And is it the right thing for geissinger in the future? Because part of modernizing our infrastructure is also modernizing our application portfolio.

So the next two slides are just going to give you a little glimpse into what geisinger overall infrastructure looked like prep project. So this is one of these exact same slide as last year. This is where we started.

Ok. So at the end of the day, most of our servers were virtualized. So we were a heavily heavily virtualized environment which set us up for a lot of this stuff as well because you virtualize one, you virtualize the next.

So a lot of concerns when you're um a major physical hardware type shop and doing physical to virtual conversions, things like that. Um I wouldn't say didn't come into play, but it was less challenging. We also we're a heavy windows shop and then the alternative for that is linux as well. So our two main um operating systems function well in almost any public cloud. So we were set up very successfully there.

Again, x 80 x 64 architectures, we had a good healthy divide between production and non production. And overall, you'll see some of the distributions there across the different things.

This one is the interesting slide for me and this is one from an automation perspective that i would say we've barely scratched the surface on so far, but are doing analysis on all of those servers in the previous slide that were in scope for migration, the applications that run on them, we ran and looked at ok. What's the cpu utilization? What's the memory? What's the dis how much technical debt is sitting there? Right? How much stuff did we overbuy? Because a vendor x said you need this or vendor y said you needed that.

So we actually ended up getting the data for this and it was a compelling part of the business case that as we get to the cloud and as we optimize, which is one of our big focuses for 2024 is to start looking at these systems and starting to actually customize and ratchet down certain resources that aren't actually needed independent of what the system requirements of an application says.

And we'll have an example of this when we talk about our e hr that we migrated to aws here in september about how being in aws allowed us to be agile with that deployment. And also with the upgrades that we've done post move have been able to actually reduce compute and not have technical debt in our data center.

So as any health care organization, we have one of the, you know, one of the popular electronic health records out on the market. Um we were one of their very first customers and this kind of sparked one of the business cases for change. And again, for us cloud isn't solely a play on cost savings as an organization, ok? Some things you put out there will cost you less, some things you put out there will cost you more. But ultimately, when you look at these, you have to look at more than just cost but agility time to deliver and all of those other things that you get.

So when we're looking at these things and our spend and everything else with the cloud and on prem, we were able to take our electronic health record over the course of this project. We had multiple environments and we started moving with our training environment first, right, the one that everybody plays on. And as part of that, we did extensive testing with the e hr vendor as well as our partner. It was deloitte and they brought some tools to bear.

And ultimately, we were able to show that performance on aws exceeded many of the performances we had on prem. Now, now part of that was when you think about your spending and being physically responsible, we knew this was coming, we knew we were doing this. So we didn't replace systems on prem. That should have been life cycled because there was no need to.

So what was really good with that is training went well. So we said, great, let's move on. Then we put another environment out there that's closer to production. We did a bunch of testing and worked great. Here are some bumps along the way. Each time we did it, we learned something new to apply to the next one.

Then we got to the point where our last environment before our disaster recovery and production, we put that out there, it continued to perform very well. So in september, we had planned to move our entire organization from operating on our e hr on prem in central pennsylvania to aws in virginia and run it for one week so that we could see performance. What do we need to do to be better at it for the big go live here at the end of the year? What did we miss in our prep? What did we miss in our steps? Right. How do we make sure that we reduce the impact to the users when we, when we finally pull the trigger here next week?

And we put it out there, we had a couple of bumps, but the issues list was much shorter than anybody anticipated. And ultimately the biggest issue that we ran across was the wrong license was applied to some of our virtual gear that limited the interface engines from being able to send their stuff as quickly as possible. So they backed up once that was identified and remediated, they flushed out in a matter of a couple of hours and about 23 days into it, it wasn't the executives, it was the engineers, it was the application owners. It was the end users of the organization that started saying, well, this is working better. It's faster. Do we really have to go back?

So there was many people that sent me emails, hit me up on teams talking to other executives that said, do we really need to move this back? So we sat down wednesday after the go live from the previous friday and had that discussion and we made the decision to leave it there. There was risk, we're now operating in a hybrid environment with our production e hr in the cloud and our disaster recovery on prem on completely different infrastructure, completely different model.

And we've been running that way since we went through an upgrade of the e hr since this, we upgraded both on prem and in the cloud with minimal bumps. We also implemented new features of the e hr that required many new servers since then in the cloud and on prem with very few bumps.

And on top of that one of the great benefits that we've realized as part of this is when we did those upgrades, these systems, typically when you have to do an upgrade or you have to add a module to it. It's, you need these couple of 100 servers for this. You need these couple of 100 servers for that. And if you don't have it, you go out, you buy it right, you bring it to the data center, you stand it all up and then you do your upgrade.

In this case, we overshot from a citrix perspective by i think a couple 100 servers, the guys in the back can keep me honest. But a lot of servers that typically we would have virtualized, they would have been run on vmware, all that great stuff. But because we were on the cloud, we just turned them off. All right, we didn't pay for them anymore. They were gone.

We had another instance again with the e hr where with the uh the big uh module we implemented, they started seeing that they didn't need all the servers based on performance metrics. So i think for about a week, they started turning off 20 servers at a time until they found the right the right thing. Historically, we would have bought the hardware for those. It may not have been much because we were heavily virtualized. But still it would have been sitting there in my data center depreciating for five years. Heating cooling everything else.

And in this particular case, we were able to just turn it off, take it out no technical debt left behind.

We're also taking this journey with our claims system. So that's about 70% out on aws right now. It's open enrollment season. So we're obviously not moving production right now, but uh that's slated to be done here in the uh february march, april time frame

"But the one point that I kind of jumped over is by us leaving it out there, the teams calculated we saved over 1000 hours of human capital by leaving it there instead of bringing it back. We've also prevented two additional downtime to our end user community because it, it took us 3.5, 4 hours to flip the order from on prem to eHR in the cloud. And during that time, things were up and down. So we didn't have to do that again.

So next week and for the rest of this year and the beginning of next year, our final eHR environment will go in our second region of Amazon and then we will start performing DR testing, flipping between multiple regions of Amazon instead of flipping on prem to the cloud and back. So with that, we've been able to minimize impact on the end user community and also saved countless hours of IT as well as everybody involved in testing.

This slide highlights many of the Amazon native services that Geisinger has used and implemented. This isn't all encompassing, but it's also meant to kind of spark the conversation of, you know, did we do it all native? And the answer is no. So as far as the way our architecture was, and the one thing that we did, which I think maybe unique in a way, especially in our industry is when this project started, we greenfield it, it was blank sheet of paper.

So we have a combination out there of infrastructure service. So your lift and shift, right? And we've got product based applications that we've built on AWS as well. So when we thought about how do we support what I'll consider a traditional IT architecture on top of the cloud as well as thinking about adaptability, innovation and everything else. We need a product based model as well.

So when we did that, we built the network design brand new, right? We interjected colo location centers to be able to have dedicated um connections as well as diverse paths to the cloud as well as that. We even diversified the technology that's used on those lines to get us to the colos, some fiber, some not different geographic regions for the paths of the cables as well as everything else.

So we took the opportunity to look at this from the time it leaves our physical facilities to Amazon and back. And then when we talk about security and a data center that you've had for many, many moons, you've been building security on top of the data center for many years, putting a little bit more here, a little bit more there, but this is brand new.

So we took a an approach, being able to separate out business units at a network layer production, non production, putting more guide rails and protections in between those things to help protect these applications. We used a lot of these native services but we also worked with some of our key partners.

So our Amazon or a AWS instance and environment is not just strictly this, we, we brought with us, you know, top grade firewalls that and part of the reason we did this and things like Citrix and F5 technology and others is we already have those skills. We have the management interfaces. So part of that was to be able to blend the new native with the technology that we had before so that when we're skilling and we're looking, we're able to tackle this all at once.

So, so with that, we're also in the constant cycle of ok, well, we've used that, can we go to a cloud native at some point? What are the things we need that may not be there? And having those conversations again with our integration partner with our AWS technical teams, things like that.

So again, a lot of this, we've used a lot of this has been extensively leveraged and some of this is on the horizon for us as well as we continue to move down this path, some lessons learned across all three of our transformation projects. And I highlighted one in each column as kind of the ones that stood out as we were thinking about this as an organization.

And when you think about your operating model of governance and partnering across the organization, promoting automation and self service to increase productivity and speed is critical. And we're laser focused on that. We haven't gotten there yet. We do a lot of automation within our data center. We do a lot of automation within the AWS environment. But that's one of our big priorities for 2024 is to continue to build our tool set out, continue to skill. Our teams continue to take traditional infrastructure and think differently about it, infrastructures, code application, developer mindset version and control, being able to recreate things with scripts instead of the traditional way of having to go through screens or rack and stack a new piece of hardware, things of that nature.

The other thing with the app rationalization is to drive enterprise standardization and consolidate redundant applications. So we'll have an analogy later. But ultimately, you don't want to take stuff to the cloud. First of all, it won't run in the cloud. And secondly, you don't want to take stuff to the cloud that you know, at the end of the day, you're not going to need more or it's going to be shut off or it's going to go away or you have four of the same things with five, with four different vendors.

So early on, this was one of our critical success factors was to really step back and look at what are the applications in our portfolio and how do we, how do we handle those? And then on the infrastructure side, the largest thing from the lessons learned and probably the biggest thing that I would encourage everybody to think about is how are you going to do organizational change management? That's not IT change management where I'm going to change this. Config this is how do you change your organization to think differently for our project?

This project was sponsored at the top level of the organization. So this had top down approval, top down, buy in. And it really allowed us to change the way that we operate, look at the way we deliver services and really engage with our user community because I wasn't joking at the beginning of the presentation when we did this project. As we spun this out, there was so much testing and so many people in the organization involved, we host our electronic health record for community partners. They were involved, they, they're not even our employees, they were involved in testing.

So this project spanned a very large, large set of team members and colleagues. So you got to make sure you're investing in the organizational change, you're investing in the skilling and the education to make your team successful. Because again, you've got a lot of um smart, adaptable resources within your organization. And, you know, it brought teams together that had never worked before together.

So I'm going to ask Angela to come up the stage. Ok? Is that not amazing hearing that story about what they've done at Geisinger? I think Dan deserves a round of applause for sharing what they've been doing. Oh, it's really for those guys in the back and everybody else that, like you said, you've got the leaders and then you got the people that actually do it. The, those those four guys back there are some of the guys that knocked us out of the park as well as our partners up front with Amazon and Deloitte. Absolutely. A round of applause for everybody. That's why it was so important for us to know who was in the room. It's the builders, it's the partners. Um it's all of you who are making a difference because we really do think this a new day in health care.

It is time to start looking at this a little bit differently and it's also time to start saying we can really get to that or suite in the future. So Dan and I wanted to talk about what is it that you need to do on Monday to get started or to advance what you're already doing? Yeah. So this is the application rationalization piece. All right. Think about getting your house in order. Are you and, and, and the operating model? Are you structured in a way? Is your organization ready to change the way that it's done. It's definitely, especially if it's in an industry or an organization that's very traditional and I don't mean traditional IT, I mean, traditional business models, traditional, you know, project management, all of those kind of things you have to think about. Are you ready to do that?

And then you have to start thinking about what your tech debt and your application debt because I've moved only a couple of times, many of you have probably moved a lot and we always pack stuff in boxes, take it with us and then never look at it again. So the idea with this and this analogy is don't pack stuff in boxes that you don't need, get rid of it because we're all going to fill our basements up with those boxes and eventually go back and wonder why we even brought it to the next house.

Now, that's a great point. The next is sometimes it can be overwhelming within an organization to say we want to do something new, whether you're gonna migrate for the first time or whether you're just gonna modernize or you're continuing to advance your current infrastructure. So what are your strategic goals? That's a great place to start. So you can look at which resources you're gonna utilize and prioritize where you're going. Yeah.

And as far as your 6, 12 and 18 month plan, there's a couple of key things that you need to think about and one of them is, are you able to do this yourself? Right. Um we had a lot of smart, capable people, some with Amazon experience because we dabbled it with it, with research and other things in the past. But we didn't have the amount of labor needed to do a project like this, right? In some cases, we needed some skills added in order to do it. So you need to think about, do you need a trusted implementation partner to help you and your team and your organization get there? Do you want to accelerate faster? Things like that?

The other piece is do not discount planning? Ok. Even though we're talking about being agile and fast and speed the delivery and all that don't rush through your design of your initial environment. Um that probably should have been one of our lessons learned. We had things that we had to go back and rearchitect because certain little pieces might have been rushed. So part of our lessons learned and things there is take the right amount of time for your organization to design your foundation because you're not going to want to go back and blow it up and redo it again.

And there was a significant amount of time spent this week on what's next, what is the emerging technology you're gonna wanna use? So now is the time to start thinking about that as we look at those challenges that were discussed earlier about the workforce. Part of the, the stress on the workforce is sometimes we're in a hurry up and get it done mode and healthcare organizations by having a plan following that path and knowing where you wanna go with the merging technology, then you're taking the whole organization along and you heard from Dan, how important that is.

And the final piece, I think I've beat this one into the ground during the presentation, but upskill your people, they're smart, they're hungry. So get them the training and the skills and everything. They need to take you into the future of where you need to go. Um because there's a lot of talent. I know there is in our organization and again, it's not just IT training, you need to educate the workforce and the executive suite and others on what this really means.

So, thank you all. Thank you Dan for telling us your story and sharing it with us. It is a delight to have you."