Latest / Becker’s Healthcare Podcast / AI, Informatics, and Value-Based Care Transformation with Edward Sankary

Transcript

AI, Informatics, and Value-Based Care Transformation with Edward Sankary · Becker’s Healthcare Podcast

• 0:00
  - This is Laura Dirda with the Becker's Healthcare
• 0:02
  - podcast.
• 0:03
  - I'm thrilled today to be joined by doctor
• 0:05
  - Edward Sankari,
• 0:06
  - vice president and chief health informatics officer as
• 0:09
  - well as chief value officer at UT Health
• 0:11
  - San Antonio
• 0:12
  - and president
• 0:13
  - of the regional physician network ACO.
• 0:16
  - Doctor Sangre, it's a pleasure to have you
• 0:18
  - on the podcast today.
• 0:19
  - Good morning. Pleasure to be here, and thanks
• 0:21
  - a lot for having me. This should be
• 0:23
  - I'm looking forward to this. This should be
• 0:24
  - fun.
• 0:25
  - Absolutely. It'll be exciting conversation. I know there's
• 0:28
  - so much happening in health care right now,
• 0:29
  - and, truly, you've got some very unique perspectives
• 0:32
  - as a clinician on the informatics side and
• 0:34
  - then the ACO as well. So I'm excited
• 0:36
  - to dig in. But before we do, can
• 0:38
  - you tell me a little bit more about
• 0:39
  - yourself as well as the organization?
• 0:42
  - Absolutely. So I'm an internist. I actually went
• 0:45
  - to medical school at the University of Texas
• 0:47
  - Health Science Center out here in in San
• 0:49
  - Antonio.
• 0:51
  - And in in college, I graduated with a
• 0:54
  - degree in computer science, and after that worked
• 0:56
  - at IBM in in New York in their
• 0:58
  - data systems division. Always had an interest in
• 1:01
  - medicine and and the
• 1:04
  - the nexus of medicine and computer technology. I
• 1:07
  - actually did some work in robotics
• 1:09
  - at IBM and out at DuPont and and
• 1:12
  - in college and ultimately
• 1:14
  - decided to go to medical school here at
• 1:16
  - UT Health San Antonio, became an internist, practiced
• 1:18
  - in Chicago
• 1:20
  - for a while, became a chief health information
• 1:22
  - officer at one of the large
• 1:25
  - community health systems that had an association with
• 1:28
  - Loyola University Health System work, which is where
• 1:30
  - I was at.
• 1:31
  - And,
• 1:32
  - ultimately, as this job came along at UT
• 1:34
  - Health San Antonio, it seemed like a great
• 1:36
  - opportunity to come back to Texas. So
• 1:39
  - moved out here. And what my role is,
• 1:41
  - it's really at that intersection of clinical informatics,
• 1:45
  - the electronic health record, third party clinical technologies,
• 1:49
  - digital transformation,
• 1:50
  - artificial intelligence strategy, and
• 1:53
  - and value based care. The University of Texas,
• 1:57
  - San Antonio,
• 1:59
  - is the largest academic medical center in the
• 2:01
  - South Texas region. We've got a large ambulatory
• 2:04
  - presence, about a thousand physicians, more than 20
• 2:07
  - ambulatory locations
• 2:09
  - throughout the region,
• 2:11
  - including a few very large multi specialty
• 2:14
  - medical centers. And
• 2:17
  - we've had a long close partnership with our
• 2:19
  - county
• 2:20
  - affiliate, University Health, where many of our clinicians
• 2:23
  - work, and and in fact, I did my
• 2:25
  - training.
• 2:26
  - But, in December
• 2:28
  - 2024,
• 2:29
  - we built our first hospital, the UT Health
• 2:31
  - Multispecialty
• 2:32
  - and Research Hospital, which we'll talk about a
• 2:35
  - little bit later. And that's really was built
• 2:37
  - as a hospital looking forward to the use
• 2:39
  - of some of these technologies.
• 2:41
  - We have a school of public health, the
• 2:43
  - Kate Marmion School of Public Health, which I
• 2:45
  - have an appointment, and and that's trying to
• 2:47
  - leverage technology
• 2:48
  - in some of the rural areas of Texas.
• 2:51
  - And just recently,
• 2:53
  - the University of Texas Health Science Center merged
• 2:55
  - with the University of Texas San Antonio, which
• 2:58
  - was our academic sister to become UT
• 3:02
  - San Antonio, which is the third largest research
• 3:04
  - university in the in the great state of
• 3:06
  - Texas. And, actually, right now, we're working very
• 3:08
  - closely
• 3:09
  - with the newly formed College of Artificial Intelligence,
• 3:12
  - Cyber, and Computing with UTSA
• 3:14
  - to to see how they can get integrated
• 3:16
  - into to our health system. And, of course,
• 3:19
  - we're part of the University of Texas System,
• 3:21
  - which has eight health related institutions and is
• 3:23
  - one of the largest
• 3:25
  - academic health systems in the country. And we're
• 3:27
  - looking also how we can leverage we can
• 3:29
  - work together
• 3:30
  - to leverage the capabilities of all of our
• 3:33
  - health related institutions
• 3:35
  - to promote art the use of artificial intelligence.
• 3:37
  - And one thing I do wanna mention, the
• 3:39
  - mission of the University of Texas San Antonio
• 3:42
  - is to make lives better by advancing
• 3:45
  - excellence in education,
• 3:46
  - research discovery, commercialization,
• 3:48
  - health, public service, and community engagement, and that's
• 3:51
  - really the mission that drives us throughout all
• 3:53
  - of our endeavors.
• 3:55
  - So that's
• 3:56
  - a a a little background.
• 3:58
  - Oh, that's really helpful to understand exactly the
• 4:01
  - position that you're in and then looking ahead
• 4:03
  - to the future, where you're going, how you're
• 4:05
  - growing. And it it seems very exciting time,
• 4:08
  - to be in San Antonio in Texas because,
• 4:10
  - you know, the technology that is coming in
• 4:12
  - is great. There's so many possibilities and potential,
• 4:14
  - but at the same time, being able to
• 4:16
  - close those gaps in the rural areas as
• 4:18
  - well,
• 4:20
  - is so critical.
• 4:21
  - Now from your perspective, what are some of
• 4:23
  - the powerful forces that are shaping digital health
• 4:25
  - and IT at the health system today, and
• 4:27
  - how do you anticipate they're gonna evolve in
• 4:29
  - the next year or so?
• 4:31
  - That's that's that's a great question. And I
• 4:33
  - usually try to frame this in a few
• 4:35
  - areas. First,
• 4:37
  - and and the most important is artificial intelligence
• 4:40
  - is moving very quickly now from experimentation
• 4:44
  - to operations. You know, a few years ago,
• 4:47
  - most organizations were just trying to pilot things,
• 4:50
  - pilot AI
• 4:52
  - use cases.
• 4:53
  - Now these tools are showing up
• 4:56
  - throughout our entire workflow spectrum
• 4:59
  - and and really integrated into what we're doing.
• 5:01
  - And we've had a number
• 5:03
  - of impactful
• 5:05
  - AI use cases that that's beginning to expand
• 5:08
  - dramatically. So
• 5:09
  - as we look at what's beginning to shape
• 5:12
  - the digital force
• 5:14
  - forces and digital health, one,
• 5:17
  - we need to understand our partnerships with industry
• 5:20
  - because that's a very powerful collaboration
• 5:23
  - to expand
• 5:24
  - artificial intelligence. One of the first ones
• 5:27
  - that many health organizations
• 5:30
  - rolled out that was
• 5:31
  - large scale and very impactful was Ambient Voice
• 5:34
  - Recognition.
• 5:36
  - Ambient Voice Recognition,
• 5:38
  - for those who may be not may not
• 5:39
  - be familiar is where a physician can come
• 5:42
  - into an exam room,
• 5:44
  - use their mobile platform.
• 5:46
  - It listens to the conversation
• 5:48
  - between the physician and the patient.
• 5:51
  - And at the end of the visit,
• 5:53
  - a clinical note is created in the format
• 5:56
  - that we as clinicians
• 5:57
  - would want it
• 5:59
  - and captures the essence of the
• 6:02
  - conversation,
• 6:03
  - helps tee up some important information.
• 6:05
  - And that's been a dramatic, really revolutionary
• 6:08
  - change for physicians because
• 6:11
  - many physicians
• 6:12
  - wouldn't be able to complete their notes at
• 6:14
  - the end of a
• 6:15
  - visit with the patient, and they'd spend what's
• 6:17
  - called pajama time. They'd spend their times at
• 6:19
  - night, on the weekends, away from their families,
• 6:21
  - trying to complete these notes. Now
• 6:23
  - when you're done with the visit, the note's
• 6:25
  - done. And it's directly improving
• 6:28
  - the the burdensome impact that actually the electronic
• 6:31
  - health health record
• 6:33
  - created for physicians. So this was revolutionary.
• 6:37
  - At at the same time,
• 6:39
  - we're working with other partners,
• 6:41
  - such as Qualified Health to
• 6:44
  - deploy
• 6:45
  - an enterprise wide secure access to the generative
• 6:49
  - AI
• 6:51
  - large language models. So now
• 6:53
  - our
• 6:53
  - any member of our health organization
• 6:56
  - can get into a secure portal
• 6:58
  - and access the large language models like Cloud,
• 7:02
  - ChatGPT,
• 7:03
  - Gemini
• 7:04
  - securely in a fashion that's safe for
• 7:07
  - protected health information, what what we need for,
• 7:11
  - the use of of including patient information in
• 7:14
  - in any of these so that patient information
• 7:16
  - never escapes beyond our secure
• 7:18
  - walls, and they can actually develop agents to
• 7:21
  - improve
• 7:22
  - workflows
• 7:23
  - in this in this platform. So, again, that
• 7:25
  - that partnering with industry.
• 7:28
  - Of course, the electronic health record
• 7:30
  - is a critical part of a of a
• 7:32
  - health system. We have Epic as our electronic
• 7:34
  - health record,
• 7:35
  - and
• 7:36
  - Epic is embedded throughout the entire clinical workflow
• 7:41
  - or operational workflows, of course, the patient workflows.
• 7:44
  - And Epic now is very,
• 7:47
  - advanced in its journey in the inclusion
• 7:50
  - of artificial intelligence
• 7:51
  - through agents. And, again, agents are
• 7:54
  - activities of artificial intelligence
• 7:57
  - that they're able to to perform and carry
• 7:59
  - out task for us that were
• 8:01
  - predesigned. And Epic's got three
• 8:04
  - agents that they have, what's called Emmy, Penny,
• 8:06
  - and Art. Emmy is the one that is
• 8:08
  - on the patient patient side. So if patients
• 8:10
  - want to
• 8:11
  - ask a
• 8:13
  - for an appointment
• 8:15
  - with their MyChart or actually on their mobile
• 8:18
  - platform in natural language, they can ask that
• 8:20
  - question, and the system can help them schedule
• 8:22
  - that appointment as opposed to going through a
• 8:25
  - much more arduous
• 8:27
  - set of questions.
• 8:29
  - Penny is what we use for
• 8:31
  - the
• 8:32
  - revenue cycle, and then ART,
• 8:34
  - the augmented response technology, is what nurses and
• 8:37
  - physicians are using to help use artificial intelligence
• 8:40
  - to compose notes, answer questions. So the the
• 8:43
  - electronic health record is a critical part now
• 8:45
  - of our
• 8:47
  - of our technology
• 8:49
  - landscape,
• 8:49
  - and
• 8:51
  - it's also
• 8:52
  - no longer just becoming part of our future
• 8:55
  - capabilities. It's it's part of our infrastructure and
• 8:57
  - even the infrastructure, the security that we have
• 9:00
  - around our system to prevent
• 9:02
  - cyberattacks.
• 9:03
  - Artificial intelligence is embedded into that platform to
• 9:06
  - be sure what we're doing is is trying
• 9:08
  - to prevent as much cyberattacks and all the
• 9:11
  - bad guys from getting us as possible.
• 9:13
  - The second key area
• 9:14
  - is governance.
• 9:16
  - Governance is an important part of
• 9:19
  - successful artificial
• 9:20
  - intelligence.
• 9:21
  - And
• 9:23
  - as we look at the significant number of
• 9:26
  - AI applications that we have,
• 9:29
  - the ability to
• 9:31
  - accurately
• 9:32
  - determine the risk of that
• 9:35
  - AI application,
• 9:36
  - how that AI application was trained, if there's
• 9:39
  - a risk of bias, if there's a risk
• 9:41
  - of
• 9:42
  - performance decay over time. We're in the process
• 9:45
  - of implementing
• 9:47
  - an AI monitoring system and management system that's
• 9:50
  - going to allow us to be sure as
• 9:52
  - we
• 9:53
  - use and deploy and evaluate
• 9:57
  - artificial intelligence that it goes through
• 9:59
  - a a process that determines the risk and
• 10:01
  - it can continue to help monitor it. So
• 10:04
  - that's another powerful force we've gotta be sure
• 10:07
  - we're incorporating into what we're doing.
• 10:10
  - Video language models
• 10:11
  - are shaping how we're handling our inpatient care.
• 10:15
  - Every single one of our rooms at the
• 10:17
  - multispecialty and research hospital
• 10:19
  - has a video platform and a wall mount
• 10:21
  - that has a camera, speaker,
• 10:23
  - microphone.
• 10:24
  - We partnered with NASA to have this equipped
• 10:27
  - throughout our entire
• 10:29
  - hospital system,
• 10:30
  - and that's allowing
• 10:32
  - our virtual nurses
• 10:34
  - to
• 10:35
  - view patients,
• 10:37
  - communicate with patients, help patients with the intake
• 10:40
  - of clarification of their medications, with their discharge
• 10:43
  - process,
• 10:43
  - help floor nurses along their workflow,
• 10:46
  - join in if there's a rapid response or
• 10:48
  - other critical need. And ultimately,
• 10:51
  - with artificial intelligence is helping monitoring
• 10:54
  - the risk of a patient for fall to
• 10:56
  - prevent them for falls or pulling out lines.
• 10:58
  - And and finally,
• 11:00
  - what we're
• 11:01
  - what we're seeing is this move away from
• 11:05
  - a point solution
• 11:07
  - to platforms.
• 11:08
  - Artificial intelligence now as opposed to developing very
• 11:12
  - small
• 11:13
  - point solutions to this problem or this particular
• 11:15
  - use case. We're seeing
• 11:17
  - a a force that's moving them into platforms,
• 11:19
  - whether the electronic health record or video platform.
• 11:23
  - That's a that's that's a direction
• 11:25
  - that industry is going. And and finally, the
• 11:29
  - artificial intelligence data platforms,
• 11:32
  - whether it's the Microsoft Azure with Fabric and
• 11:34
  - AI Foundry,
• 11:35
  - that's another critical part of the way the
• 11:39
  - industry and digital health and IT at our
• 11:42
  - organization is is transitioning
• 11:44
  - because these platforms give us the ability to
• 11:48
  - look at data, to ask questions about data
• 11:50
  - that we've never had before.
• 11:52
  - So
• 11:53
  - those are some of the major forces that
• 11:56
  - are impacting the technology, but we can't forget
• 11:59
  - workforce development.
• 12:00
  - To successfully
• 12:02
  - use these technologies
• 12:03
  - and to use them at scale, we need
• 12:05
  - to be sure that our workforce is adequately
• 12:08
  - trained
• 12:09
  - on how to use the technologies
• 12:10
  - as well as how to safely
• 12:13
  - use the technologies.
• 12:14
  - So those are some of the major
• 12:16
  - forces that we see shaping digital health and
• 12:19
  - information technology at our organization
• 12:22
  - and and the direction that we're seeing over
• 12:25
  - the next
• 12:26
  - year,
• 12:27
  - our institution move.
• 12:29
  - That's really helpful to understand. And, you know,
• 12:31
  - looking at all those trends, obviously, moving more
• 12:33
  - to the platform, seeing how AI is integrating
• 12:36
  - strongly into the operations and,
• 12:38
  - supporting clinicians as well.
• 12:41
  - It it seems like so many moving parts
• 12:43
  - and a lot of change happening at once.
• 12:46
  - From your perspective, can you tell us a
• 12:49
  - little bit about or zero in on,
• 12:51
  - something you did in the last year that
• 12:53
  - really had great results? What did you do,
• 12:55
  - and how did you, really get there?
• 12:58
  - Absolutely.
• 13:00
  - One of the biggest
• 13:01
  - use cases that we implemented
• 13:04
  - in the past year,
• 13:05
  - was for
• 13:07
  - the
• 13:08
  - standing delegated orders. We're gonna talk about those.
• 13:10
  - So it was in the area of compliance.
• 13:13
  - In December 2024, as we said, we built
• 13:16
  - our first hospital, the multispecialty and research hospital,
• 13:19
  - and that allowed our clinics to convert to
• 13:23
  - HOPD or what's called hospital outpatient departments.
• 13:26
  - And when you make that conversion,
• 13:29
  - you need to get
• 13:30
  - a compliance accreditation,
• 13:33
  - by the joint commission.
• 13:36
  - And one of the areas
• 13:38
  - that we had that required a significant
• 13:41
  - amount of work and we had to do
• 13:42
  - this over a rapid time frame
• 13:45
  - was standing delegated orders. And these are orders
• 13:48
  - that
• 13:49
  - clinicians
• 13:50
  - have said that nurses and medical assistants and
• 13:54
  - other non licensed providers would carry out on
• 13:57
  - their behalf,
• 13:58
  - but the physicians ultimately are responsible and need
• 14:01
  - to sign for and authenticate
• 14:03
  - those orders.
• 14:06
  - And we had a whole bunch of those,
• 14:08
  - and we had to
• 14:09
  - get those into
• 14:11
  - a compliant fashion very rapidly.
• 14:15
  - So what we did is we worked with
• 14:16
  - our electronic health record
• 14:19
  - and
• 14:20
  - ensured that every order
• 14:21
  - a MA
• 14:23
  - nurse could place what's standing delegated order this
• 14:25
  - came from. And there may be a standing
• 14:27
  - delegated order of vaccinations or a standing delegated
• 14:30
  - order for x rays for orthopedic surgeons in
• 14:33
  - their clinics.
• 14:35
  - But because there's so many of these standing
• 14:37
  - delegated orders, it was obviously difficult for the
• 14:40
  - nurses to know which order
• 14:42
  - was within which standing delegated order. So we
• 14:45
  - developed an agent
• 14:47
  - along with our qualified health platform where the
• 14:49
  - nurses could query and ask if I wanted
• 14:52
  - to,
• 14:53
  - can I give an HPV vaccine to a
• 14:56
  - 55 year old male? And they would put
• 14:58
  - that information in, and the the agent would
• 15:01
  - say, nope. Can't give that. That's not meeting
• 15:03
  - current,
• 15:04
  - requirements within our institution. But if they said
• 15:07
  - for 25 year old female, would say, yes.
• 15:09
  - And here's a standing delegated order that it
• 15:11
  - came from, and the nurses could ensure that
• 15:14
  - the order that they're placing indeed came from
• 15:17
  - a standing delegated order.
• 15:19
  - On the back end,
• 15:21
  - and this was almost the
• 15:23
  - the the the most important part,
• 15:25
  - is
• 15:26
  - every standing delegated order that was placed, we
• 15:29
  - were able to determine,
• 15:31
  - did it indeed come from a standing delegated
• 15:34
  - order,
• 15:35
  - and did it indeed come from the standing
• 15:37
  - delegated order that the
• 15:39
  - nurse or medical assistant
• 15:41
  - placed in the in in that field that
• 15:43
  - asked for which standing delegated order came from.
• 15:46
  - That allowed us to achieve
• 15:48
  - 100%
• 15:50
  - compliance
• 15:51
  - with standing delegated order use, which was astronomical,
• 15:55
  - almost unheard of.
• 15:57
  - And we did this through
• 15:59
  - the use of large language models that are
• 16:02
  - looking at every single order that we placed
• 16:06
  - using,
• 16:07
  - the standing delegated order, these forms that were
• 16:10
  - created, and
• 16:12
  - ensuring that order came from the the form.
• 16:15
  - And when
• 16:16
  - the joint commission came here, they actually were
• 16:18
  - extremely impressed by the ability to ensure that
• 16:22
  - we have a 100% compliance in this very
• 16:24
  - important part of patient care. So that was
• 16:26
  - one of the big swings that we took,
• 16:28
  - you know, how could we rapidly deploy an
• 16:30
  - AI agent to ensure that we met compliance
• 16:33
  - needs for the joint commission. And we did
• 16:35
  - it in a way that ultimately ended up
• 16:37
  - improving patient safety
• 16:39
  - and improving efficiency
• 16:41
  - for our staff.
• 16:42
  - Two other areas I'd like to talk about
• 16:44
  - that are big swings that we're in the
• 16:46
  - process of of taking.
• 16:48
  - One is we're
• 16:50
  - working with
• 16:52
  - Microsoft
• 16:52
  - and
• 16:53
  - taking our
• 16:55
  - ACLs or accountable care organization platform
• 16:58
  - and moving that to an AI platform on
• 17:01
  - Azure
• 17:02
  - using Azure Fabric and Foundry, which are their
• 17:05
  - AI tools, developing,
• 17:07
  - a semantic layer so the system will automatically
• 17:10
  - understand if we ask a question
• 17:13
  - to the platform
• 17:15
  - where that information is coming from and and
• 17:17
  - help us answer this in just a natural
• 17:19
  - language query. So if I wanted to ask
• 17:22
  - a question, what's our readmission rate? What's our
• 17:25
  - readmission rate for a particular time frame?
• 17:27
  - What's our readmission
• 17:29
  - rate
• 17:30
  - for patients who have congestive heart failure. We
• 17:33
  - can ask that in a natural language process
• 17:36
  - and get a response back. And and and
• 17:39
  - with this technology,
• 17:40
  - we're feeding massive amounts of data that the
• 17:42
  - system is figuring out what that data means.
• 17:45
  - That's that semantic layer.
• 17:46
  - And finally, we're
• 17:48
  - developing as we set a collaboration
• 17:50
  - with
• 17:51
  - the College of AI, Cyber, and Computing at
• 17:54
  - UTSA,
• 17:55
  - bringing in some of the PhDs
• 17:57
  - who have expertise
• 17:59
  - in video language models and large language models,
• 18:01
  - but don't know medicine, and they're helping us
• 18:04
  - because they know the AI side. So those
• 18:06
  - are three big swings that we're taking this
• 18:08
  - year.
• 18:09
  - Oh, that's amazing to hear. And, you know,
• 18:11
  - really incredible to understand the results that you're
• 18:13
  - getting as you dig deeper into AI, creating
• 18:15
  - those agents, and,
• 18:17
  - figuring out how to make the large language
• 18:19
  - models work. I I love the fact too
• 18:21
  - that you're working, you know, with different companies,
• 18:23
  - both on the health care provider side as
• 18:25
  - well as on the ACO moving the platforms
• 18:28
  - together and,
• 18:29
  - really being able to it seems like making
• 18:31
  - things a lot more seamless and safe for
• 18:33
  - the patient data.
• 18:36
  - That's right. Absolutely. You know, that's
• 18:38
  - the you know, as as we look at
• 18:40
  - the the major problems that that we're trying
• 18:43
  - to solve right now, you hit the two
• 18:45
  - major problems, you know, making things
• 18:48
  - easier for clinicians, for the nurses, for the
• 18:51
  - physicians, and ensuring that the data that we
• 18:53
  - have is is safe.
• 18:56
  - The, you know, the the the documentation
• 18:59
  - burden for physicians had become overwhelming.
• 19:02
  - And as we've partnered with our Ambient voice
• 19:04
  - recognition partner, Abridge,
• 19:06
  - we're actually seeing that physicians are
• 19:10
  - decreasing
• 19:11
  - the amount of time that they have to
• 19:12
  - spend in notes. And as we said, decrease
• 19:14
  - in this pajama time, and that's making a
• 19:16
  - significant
• 19:17
  - improvement in the lifestyle for clinicians, but it's
• 19:20
  - doing more than just that.
• 19:22
  - It's now remarkably allowing physicians to sit in
• 19:25
  - an exam room and actually look at the
• 19:28
  - patient, talk with the patient, and not being
• 19:31
  - stuck typing on the keyboard, looking on the
• 19:33
  - screen.
• 19:34
  - And we're seeing that these
• 19:36
  - technologies are advancing to where some of that
• 19:38
  - information is gonna be able to be displayed
• 19:40
  - in screens in the exam room.
• 19:43
  - So this the the technology is really solving
• 19:45
  - that difficult problem of how do you reduce
• 19:48
  - the burden that and the complexities of current
• 19:51
  - medicine has caused for physicians,
• 19:53
  - including even the ability
• 19:56
  - to begin
• 19:57
  - to tee up the things that are needed
• 19:59
  - for prior authorizations,
• 20:02
  - so that we get the right information
• 20:04
  - at the time of the visit and the
• 20:06
  - patient can have that procedure, that test authorized
• 20:09
  - immediately.
• 20:11
  - And and we're seeing that it's also helping
• 20:13
  - with clinical decision support so that clinicians are
• 20:16
  - able to
• 20:17
  - get
• 20:18
  - information
• 20:19
  - that they need to be sure that they're
• 20:21
  - providing the right care
• 20:23
  - at the right time for the patient. And
• 20:25
  - then
• 20:26
  - safety is the the other key area. We're
• 20:29
  - we're we're
• 20:30
  - extraordinarily focused on being sure that our systems
• 20:33
  - are extraordinarily
• 20:35
  - safe and that we're following a governance process,
• 20:38
  - and then finally, that we're improving our our
• 20:40
  - care delivery.
• 20:42
  - Those are areas that are absolutely,
• 20:44
  - the major importance that we're trying to solve
• 20:46
  - with AI and technology. And I'd add on
• 20:48
  - the the final one, as we touched on
• 20:50
  - before,
• 20:51
  - is our ability to leverage these powerful platforms
• 20:55
  - to allow us to extract the maximal insight
• 20:58
  - that we can from data. You know, health
• 21:00
  - care is the largest source of data in
• 21:02
  - the world and and continuing at that rate
• 21:05
  - to
• 21:06
  - exceed all other sources of data.
• 21:08
  - With AI, we're finally able to leverage this
• 21:11
  - massive data to be sure that we're providing
• 21:13
  - the best care for the patients.
• 21:16
  - It makes a lot of sense. You know?
• 21:17
  - And it's really a helpful oversight to,
• 21:20
  - look at where you're at today and and
• 21:21
  - some of the big challenges in health care
• 21:23
  - that you're able to tackle because of the
• 21:25
  - technology.
• 21:26
  - Now before we wrap up here, I just
• 21:27
  - have one more question. What are you most
• 21:29
  - curious about for the future?
• 21:32
  - Right. That's a tremendous question.
• 21:36
  - The the things that we're most curious about
• 21:38
  - for the future
• 21:40
  - is
• 21:41
  - a few areas.
• 21:43
  - And one of them is
• 21:46
  - how we're going to ensure that our workforce
• 21:49
  - evolves.
• 21:50
  - This technology is is obviously extremely powerful,
• 21:54
  - but along with it, we've gotta be sure
• 21:56
  - that the workforce
• 21:58
  - evolves
• 21:58
  - with the technology. So we've gotta continue
• 22:02
  - to watch that, help direct direct that, help
• 22:05
  - ensure that we're providing the right education for
• 22:08
  - our workforce, our staff to be sure that
• 22:10
  - they're they're maximizing
• 22:12
  - the capabilities
• 22:13
  - of this platform.
• 22:15
  - Two is the the partnerships
• 22:17
  - that are are forming.
• 22:19
  - You know, we're
• 22:20
  - as we discussed, we're forming this partnership with
• 22:23
  - UTSA,
• 22:24
  - with the College of
• 22:26
  - AI, Cyber, and Computing that's bringing in this
• 22:29
  - expertise and allowing us to maximize the capabilities
• 22:31
  - of AI in the health system. How these
• 22:34
  - partnerships are gonna further develop even at the
• 22:37
  - University of
• 22:38
  - Texas system level, the health system level,
• 22:40
  - we're somewhat novel in that all of the
• 22:43
  - UT health systems are working together in a
• 22:46
  - initiative called UT Real Health AI, the UT
• 22:50
  - Research Engineering Application Laboratory for Healthcare Artificial Intelligence
• 22:54
  - that Doctor. Zerwas, our chancellor, who was the
• 22:56
  - vice chancellor of the health related entities at
• 22:58
  - the time, helped sponsor.
• 23:00
  - And that's bringing together the entire UT Health
• 23:03
  - system, all HRIs, and seeing what can we
• 23:05
  - do if we work together to solve these
• 23:08
  - problems.
• 23:09
  - The
• 23:10
  - technologies
• 23:11
  - that that are
• 23:13
  - out there now, like, that are embedded in
• 23:15
  - our workflows,
• 23:16
  - that's gonna be curious to see the direction
• 23:18
  - that that goes. The electronic health records are
• 23:21
  - beginning to include so much technology and clinical
• 23:23
  - decision support.
• 23:25
  - You can only imagine where it's going. One
• 23:27
  - of the the the areas that Epic has
• 23:30
  - has been working on is a medical events
• 23:33
  - model that's actually able to begin to predict
• 23:37
  - medical events for the future, kinda like large
• 23:40
  - language models, predict the next word at the
• 23:42
  - AskCHAT GPT. This is gonna begin to predict
• 23:45
  - what the next test is, what the next
• 23:48
  - potential illness is to allow us not only
• 23:50
  - to help reduce the consequences of some of
• 23:53
  - these illnesses, but maybe to prevent them.
• 23:56
  - And then finally, you know, we need to
• 23:58
  - look at where
• 24:00
  - large industries going. Yeah. We've got
• 24:03
  - Epic as an electronic health record that's embedding
• 24:06
  - massive AI into their platform.
• 24:09
  - We've got the other technologies like Abridge or
• 24:12
  - Ambient Voice Recognition that's beginning to include
• 24:15
  - clinical decision support directly in line with what
• 24:18
  - they're we're doing. We've got the hyperscalers that
• 24:21
  - are working to
• 24:22
  - reduce the cost of tokens,
• 24:24
  - the development of these large language models, OpenAI,
• 24:28
  - Anthropic.
• 24:28
  - NVIDIA
• 24:30
  - is obviously
• 24:31
  - in in the space looking at how can
• 24:32
  - they
• 24:33
  - address, you know, this massive data on imaging
• 24:36
  - and clinical health care and begin to put
• 24:38
  - this all together to improve health care. And
• 24:41
  - finally, all the
• 24:42
  - the information
• 24:44
  - retrieval and clinical decision support companies, open evidence
• 24:48
  - up to date, you know, how they're beginning
• 24:50
  - to
• 24:51
  - interact
• 24:52
  - with clinicians to ensure that we get the
• 24:54
  - right information that we need rapidly.
• 24:57
  - And the final area I would mention
• 24:59
  - is how ultimately
• 25:01
  - patients are gonna begin interacting
• 25:04
  - through these AI technologies
• 25:06
  - with their health data.
• 25:07
  - It's probably gonna be moving away from,
• 25:10
  - just a portal
• 25:12
  - that they log into and search around to
• 25:14
  - find information,
• 25:15
  - to being able to ask questions
• 25:18
  - in an AI platform,
• 25:20
  - with conversational
• 25:21
  - AI about their health care and get the
• 25:23
  - right information that they need and ultimately help
• 25:26
  - their movement through the health system. So that's,
• 25:29
  - as I see it, the major curious points
• 25:31
  - that we'll be looking at over the next
• 25:33
  - few years.
• 25:35
  - I love it. That's amazing to hear. Doctor
• 25:37
  - Sankari, thank you so much for joining us
• 25:39
  - on the podcast today. This has been such
• 25:41
  - a a fun conversation. I think we covered
• 25:43
  - a ton of ground, but so inspiring to
• 25:46
  - think about what's possible today and then where
• 25:48
  - the field is headed. And I know we'll
• 25:50
  - dig deeper into all of these themes at
• 25:52
  - the health IT and revenue cycle conference coming
• 25:54
  - up in September. You'll be speaking there, and,
• 25:57
  - it'll be truly amazing to have you and
• 25:59
  - many of your colleagues be part of this,
• 26:01
  - incredible event. So I'm looking forward to seeing
• 26:03
  - you there.
• 26:04
  - Absolutely. I wanna thank Becker. It's really amazing
• 26:06
  - that you're that you're doing this. This is,
• 26:08
  - you know, very important for
• 26:10
  - for health care, very important that we get
• 26:12
  - this information out. I I thank you for
• 26:13
  - the opportunity to speak and look forward to
• 26:15
  - seeing everybody in September. Thanks again.