This Week Health 5 Years


December 8: Today on the Community channel, it’s an Interview in Action live from CHIME with Christian Carmody, CTO & SVP at UPMC. Does being a leader in the field and always trying to stay on the cutting edge create its own set of challenges for UPMC? How are they utilizing AI and machine learning? How are they using data to address some of the social determinants of health challenges for the population they serve?


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

interview in action from the:

All right, here we are for another interview in action from the Chime Fall forum, and we're here with Chris Carty, the CTO for U P M C. Chris, looking forward to the conversation.

Likewise, bill. Thanks for having me.

You know what's interesting, you're the first person from U P M C I think I've had on the show in five years outside of Rob Demic.

Okay. But by the time he was on the show, he was.

Well, I don't think I'm retiring anytime soon. So it's, it's great to hear and great to represent U P M C.

Rob and I now have an annual show that we do together Oh. Where we look at the healthcare finances. Oh, great. And he's such a great, has such great insight being former cfo.


Just very, very great guy and very knowledgeable and, very forward thinking.

So, cto, like, what's, tough in mind? What are you, what's going on? What are you excited?

Uh, There's, there's a long laundry list, but at the top of that laundry list are things like our clinical analytics and, how we're using data to unlock , and learn new ways to care for patients.

So that's, probably number one in using technology to do that. We have such a difficult time trying to acquire data from these legacy electronic health record systems and ancillary systems and really other sources of data. Whether it's, their interactions with our call center.

From a patient perspective, we're trying to ingest all that data and make some sense of it and how we can better, sort of skate to where the puck is heading. Mentality with our analytical


is the, it's interesting. U P M C is often considered a leader, especially this data space. In fact, if I go back to when I was a CIO at St.

Joe's, we reached out to U P M C and got our data teams together. You guys were always. Cutting edge of, different data platforms and those kind of things, does that create its own set of challenges being trying to be


It makes, it makes it interesting and it's, it's really about, pursuing that next learning with how we, again, deal with our patients and we're driven by our great clinical leadership and, researchers that are, you know, exploring those new learnings, which is ultimately gonna impact how all the patients, all the members of our health plan and all the communities that we.

From a technology perspective, it's always great cuz you get to work on the, newest things, newest technologies and apply that to our spaces , in healthcare.

So you say newest technologies. I immediately go to ai, machine learning and those kinds of things. How are you tapping into that view of platform?

Are, are you. Utilizing cloud vendors or are you building stuff OnPrem?

More likely moving to the cloud. We're actually re-platforming our existing clinical analytics, which has been in place for about 10 or 12 years and has done a great job, but we've hit up against limits from a data perspective, from a performance perspective.

And our goal is really to gather those insights at the point of care. So really excited about what we're able to do with re-platforming our data analytics to cloud-based technologies and using machine learning, artificial intelligence to help speed up and accelerate how we get those learnings, those insights back at the point of care.

So our doctors and nurses that are looking at a patient face to face have those. At their disposal so that can guide their clinical decisions.

Are we, man, I have like 50 questions I wanna ask. That's, I'm gonna have to squeeze this into 10 to 12 minutes. Are we applying AI and ML to the data quality at this point?

Both. It's, not only just to acquire clean data or to reconcile that data, but also then to,

oh, so you're applying it at the point where it's coming into the system. Clean up

just to acquire it. , it's a difficult hurdle. About 50 to 60% of our staff's time and effort is spent on acquiring the data.

So it's taking outta some legacy system and some legacy format and trying to massage it, clean it up, and bring it into a useful format that we can apply ML AI to too.

So you, you're also taking on the challenge of getting it back into the workflow and you guys are a epic. Cerner chef,

all the above.

Yeah, I, through acquisition, organic growth, , we've have two versions of Epic. We have two versions of Cerner, and we have about five other EHRs through acquisitions of some of the hospitals that we've acquired in the last few years.

So did you have to create another layer to deliver it?

It's not really another layer.

It's really just another interface, another data acquisition strategy to extract that data. , as quickly as possible. And as you know, it's very difficult to get that data in a timely manner.

That's interesting. So delivering it into the workflow, so you guys are also academic. You're doing a lot of research, getting, I assume a fair amount of grants and those kinds of things around that.

How much progress have we made in terms of the, the self-service? Platform where people are, I mean, where the clinicians are actually starting to do their own studies on the data.

We've actually been doing that for surprisingly, a long time and working with our researchers and clinicians about serving them up the data.

The challenge now though, is they want more data and they want it faster, and that gap we're trying to close with replatforming to the cloud and using ML and AI to help us get there faster.

Is is the challenge. You know, It's the usual thing with data the size of the amount of data you're talking about.

At this point,

It's the amount of data and it's the different types of data. So a great example, we're working very closely with our ophthalmology department around digital twin and creating synthetic data that we don't have on these patients regarding social determinants of health, but generating that data so they can explore different clinical pathways.

And so it's, sorta dabbling in that metaverse world that, we all find very interesting.

Think about the metaverse. You have this, this persona out there, so you apply it to patient care.

You can go down a clinical pathway and explore from a risk perspective or just from an outcomes perspective. What pathway is best for them? So our ophthalmology group, they're looking at how to cure blindness, right? How to treat some like, again, historical issues with, the eye.

So we're taking digital images of the retina and they're mapping that back to the cellular level, which they don't have today. So that's some of the, like the work and the research that they're doing, they're doing on a small data set. In terms of the number of patients, only probably like a few hundred at this point, but the data itself is, it's ginormous data sets.

The imaging data and the different scans and the cellular level of data. So that's that, that's a lot of our challenges and, and you can't do that on-prem. You have to use some pretty powerful computing mechanism to make that happen

📍 📍 All right. We're gonna be doing webinars a little different this year. I've talked to you a little bit about this. We got together with our advisors. They told us, Hey, you gotta do 'em different. They're just not serving the community well. And we said, what do you want? They said, community generated topics.

Great contributors. Not product driven. They want , a more honest and open discussion. And they said what we want is no on-demand webinars. We want once and done type webinars on a consistent date and time. So every first Thursday of the month. Our first one being January 5th first Thursday a month, one o'clock Eastern time.

th, priorities for:

Discussion with Integrated Delivery Networks, February 2nd, we're gonna come back with Academic Medical Center CIOs talking about their priorities. And then we're gonna hit some of the other great topics that they've given us for the year, and we would love to have you join us again this week,, top right hand corner, it'll have our current webinar and our upcoming webinars.

You can sign up right there. And if you miss it, it's not on demand anymore, so. We would love to have you there. Make sure somebody from your team is there taking notes and bringing stuff back to your staff. So we hope that this works out. Any feedback? Go ahead and send us a note. We would love to hear about it.

All right, let's keep going. 📍 📍

so that Western Pennsylvania area, you guys serve a significant portion of that population.

Talk a little bit about social determinants. I assume you're bringing in a bunch of external data sources now, trying to marry that in order. Address some of those challenges.

So again, we've,, always tried to address and gather the data at the point of registration when a patient enters our system to ask some questions and gather that data.

I don't know if we've done an effective job of using that data in our decision making. With the exception of population health, right? You can look at large data sets and this zip code, you have this many people with a particular, disease. That might be concerning, but you, have to deep dive deeper.

That deeper dive is really what we're striving towards when you talk about the social determinants health. And that gets back to how we use technology and creating some synthetic data around that to fill in those gaps. We don't know exactly what their disease state is or what their condition is regarding, like social determinants

of health.

The Nirvana state though, When the clinician is getting ready to discharge somebody, they see some of that social determinants. I remember a study one time said the number one indicator of if somebody was going to readmit was, did a family member show up to pick them up or did they sort of leave in a cab or, that kind of thing.

And if a family member, it just indicated that they probably had a good social structure to take care of them and they were going to be able to adhere to their clinical protocols.

and those kind of Things,

and that's a great example, right? And that information and data is probably buried somewhere in a clinical note or discharge summary that traditionally when you look at how analytics and data have been used in a structured format, you miss it.

So what we've done another AI platform that we used taking text based records and unstructured data and, basically structuralizing it so we can consume it. And that's definitely one of the areas that we're pursuing with a lot of, I would. say 70% of our data's unstructured and it's untapped learnings and knowledge that we have to

explore further.

And that's no small deal. I mean, it's not, we throw, oh, it's nlp. It's just nlp. You do that, but those notes are so varied and finding that data and then it has to be accurate. Right? So you're trying to pull that and move it into the discreet data elements that can actually be working in what is outside of NLP at this point.

How are you? That,

well, again, it's, taking , now structured data and rarely aggregating it, different data elements of it. So it's saying, okay, like you have this condition. You were at this hospital, you live here, any of that data aggregated together to see if there's, insights or learnings, or sometimes there won't be.

But that's part of this kind of very interesting process of exploring data and how we can apply back to providing care to our patients.

It's interesting, the, this whole conversation has been, Clinical data and the application of clinical data has really been about the power that computing can bring to delivering of healthcare.

Honestly, a lot of my CTO conversations are about, oh, hey, we're moving to the cloud and this is what we're doing on the infrastructure side. Even some security stuff falls under the uh, CTO as well.

Yeah, I, I have that responsibility as well.

How much of your time are you spending in the, data area versus the.

Really the architecture and, and that side of it.

Surprisingly I would say 20% of my time is in the architecture space and the other 80% is, again, problem solving, trying to drive up our, scalability, our resiliency as an organization, whether it's around the data piece and analytics or cybersecurity.

Obviously very hot topic, important topic to make sure that our organization and, really the challenges in healthcare of keeping our organiz. Alive and well when you're under attack or if you've been attacked from a cyber perspective.

So from a growth and acquisition standpoint, when there is new data sets to bring in talk to me about the process that you will go through, a new organization gets acquired.

You decide, it's an interesting strategy by the way. You're one of the few that's still employing the strategy of Yeah, we're just gonna leave the EHR there. But that creates an awful lot of, challenges, not only in terms of bringing the data into the analytics and the quality programs, but it's the ongoing, right?

It's the ongoing of , the quality metrics and all that kind of stuff. You almost have to create multiple workflows around that. I mean, it's, is that something you guys are looking at or,

absolutely. And we're always exploring like, what's the best way to do this? How we can maybe create a.

Adopt one platform from an EHR perspective. But you're absolutely right when you bring on a new organization, and it could be as small as one physician practice, it doesn't have to be a big hospital, , there's an integration period that culturally, this is how we do things. And trying to assimilate to how, the greater U P M C


Yeah. Cuz you're trying to drive out clinical variation, you're trying to drive higher quality scores and customer satisfaction and all those measures across multiple EHRs. And we haven't even talked about the, ambulatory side like you're talking about, where you could have not only several, you could have

hundreds, hundreds, and again, think about hundreds of different small businesses that you're bringing into your organization.

So again, to me it's more about making that, , whether it's doctors, nurses, clinicians, whomever, being part of U P M C, understanding culturally, this is how it works and, this is how you can leverage the, our size and scale of U P M. To improve how you deliver patient care and how you improve the quality of that care that you're delivering.

Fantastic. Chris, it's

great meeting.

Bill. Nice to meet you as well. Great talking with you.

Thank you.

Another great interview. I wanna thank everybody who spent time with us at the conferences. I love hearing from people on the front lines and it is Phenomen. That they have taken the time to share their wisdom and experience with the community, which is greatly appreciated. We also want to thank our channel sponsors one more time, who invest in our mission to develop the next generation of health leaders. They are Olive, Rubrik, trx, Mitigate, and F5. Thanks for listening. That's all for now.

Want to tune in on your favorite listening platform? Don't forget to subscribe!

Thank You to Our Show Sponsors

Our Shows

Keynote - This Week HealthSolution Showcase This Week Health
Newsday - This Week HealthToday in Health IT - This Week Health

Related Content

1 2 3 190
Amplify great thinking to propel healthcare forward and raise up the next generation of health leaders.

© Copyright 2023 Health Lyrics All rights reserved