Latest / Becker’s Healthcare Digital Health + Health IT / Cheng-Kai Kao, MD, FACP, SFHM, Chief Medical Information Officer at UChicago Medicine

Transcript

Cheng-Kai Kao, MD, FACP, SFHM, Chief Medical Information Officer at UChicago Medicine · Becker’s Healthcare Digital Health + Health IT

0:00
Exciting things are happening at Becker's Healthcare. Stay
0:04
ahead of industry trends with the new Becker's
0:06
CFO plus Revenue Cycle podcast,
0:08
your go to source for insights from top
0:11
healthcare finance leaders. Tune in wherever you get
0:14
your podcasts.
0:15
And don't miss the tenth annual Health IT
0:18
plus Digital Health plus RCM conference, happening September
0:22
30 to 10/03/2025
0:24
in Chicago.
0:25
Join thousands of executives,
0:27
engage with industry leaders, and explore the future
0:30
of health care innovation.
0:32
Learn more about our upcoming events at beckershospitalreview.com.
0:36
See you there.
0:38
This is Laura Dirdle with the Becker's Healthcare
0:40
podcast. I'm thrilled today to be joined by
0:42
doctor Chen Kai Kao, chief medical information officer
0:45
at UChicago Medicine. Kai, it's a pleasure to
0:48
have you on the podcast today.
0:50
Thank you so much, Laura. It was nice
0:52
meeting you, you know, virtually.
0:54
So thanks for having me.
0:56
Absolutely. Absolutely. Well, I I'm excited for our
0:59
conversation because I know you're doing some really
1:01
cool things at UChicago Medicine, and certainly, we'll
1:04
look forward to learning more about those, when
1:06
you speak at our health IT digital health
1:08
revenue cycle event later this year. But
1:11
for our conversation here, before we begin, I
1:13
was wondering, could you tell us a little
1:14
bit more about UChicago Medicine and what makes
1:16
it unique?
1:18
Of course. So the University of Chicago Medicine
1:21
health system is, you know, in Illinois, obviously,
1:26
centered in the Hyde Park area in Chicago,
1:28
but we also have
1:30
satellite hospitals at
1:32
Southwest
1:33
Suburbs and also Northwest Indiana.
1:36
We are a five hospital system,
1:38
about 1,300 beds,
1:40
and spread again across
1:42
the State
1:44
of Illinois and Indiana.
1:46
I think I would say one thing unique
1:47
in our institution is,
1:50
like many others, we are very committed to
1:52
patient care, but we do serve a relatively
1:55
underserved
1:56
population here. So there was one New York
1:58
Times article citing
2:00
the expectancy
2:02
in the zip codes that we serve
2:04
nearby
2:05
was thirty years less than the zip codes
2:07
in Downtown Chicago. So you speak to a
2:10
lot
2:11
of the resource differences,
2:14
the digital literacy, digital equity differences
2:17
and just all in all,
2:18
I think
2:19
signified
2:20
the the challenges we we should face when
2:22
we, you know, try to serve this community.
2:29
Got it. That's helpful to know. And, you
2:31
know, really an important population you serve as
2:33
you mentioned,
2:34
those who may not have access to care
2:36
otherwise, especially, and then,
2:38
truly, just a great academic health system,
2:41
resource for the Hyde Park community.
2:43
Now I'm curious. When you look back in
2:45
the last year or so, what are the
2:47
accomplishments that you're most proud of?
2:50
Yeah.
2:51
I think over the years, we we have,
2:52
you know, have I think done a few
2:55
things I think, you know, we feel very
2:56
proud of.
2:58
In my portfolio, I would say one thing
3:00
that's relevant
3:01
to what I just mentioned about serving the
3:03
community, you
3:05
know, in our area will be a hospital
3:07
home program.
3:08
So it's
3:09
it's basically a program providing acute hospital level
3:12
care in the patient's home.
3:14
And we supply, obviously,
3:16
you know, staff and resources like medications,
3:20
meals, etcetera to the patient's home. We can
3:22
do imaging and
3:25
EKGs and other studies at home as well.
3:27
And but like I mentioned, I think the
3:29
geography that we serve posed a lot of
3:32
challenges both from the patient's readiness to receive
3:35
the type of services
3:37
and also the recruitment for nurses who is
3:39
willing to travel into these areas. But we've
3:41
been fortunate to have a very strong team
3:43
where
3:44
we are really eager
3:45
to serve our patients. We
3:48
reach out to them passionately
3:49
and introducing the program. And I'm proud to
3:52
say that all the patients go through the
3:54
program is very happy with the program. They
3:56
really feel like this allowed them to be
3:58
able to recover in their own home, you
4:01
know, while, you know, really just enjoy the,
4:03
you know, all the benefit being being at
4:05
home, surrounded by, you know, your, loved ones
4:08
and also be able to do the stuff
4:10
they usually do. You know, watch the TV
4:12
show you always watch, read the books they
4:13
usually do while you're receiving all the care
4:16
that's equivalent to the inpatient level care. And
4:19
we are seeing, you know, so huge patient
4:21
satisfaction,
4:22
a lower readmission rate,
4:24
and that's just, you know, overall very fulfilling.
4:28
There's also a couple of other informatics related
4:30
projects,
4:31
that we implemented.
4:32
One thing we'll highlight is called auto diagnosis.
4:35
This is a project actually spearheaded by one
4:38
of our physician
4:40
builder, Doctor. Medcel Ocelli, really sort of building
4:43
these
4:44
auto diagnosis within the node templates to help
4:47
reduce physician documentation burden. I mean, as we
4:50
all get recognized,
4:52
the documentation is a huge
4:54
burden for the physician, not necessarily what we
4:56
signed up for medical school for,
4:59
but obviously it's
5:02
an important task that we have to do
5:03
every day.
5:04
So the the project is basically allowing us
5:07
to be able to help physician capture a
5:09
company made diagnosis
5:11
directly within their note templates
5:13
and in the meantime reducing the core inquiries
5:15
that people will be getting down the road.
5:16
So it really helped reduce the burden and
5:18
actually helped to make our notes better.
5:21
So we since they have published
5:23
the work, we actually soon to have,
5:25
additional publications on this work coming up as
5:28
well.
5:29
We have shared that in a couple of
5:30
different epic conferences,
5:32
and actually was highlighted, I think, in one
5:34
of the Becker newsletter before. And we have
5:37
been working with some of the other institution
5:39
who are interested in implementing the same,
5:43
same sort of technology in their
5:45
system as well.
5:47
Last but not least,
5:49
so clinical decision support is a big chunk
5:51
of my portfolio. So we've been using decision
5:53
support to help capture
5:56
some of the stewardship
5:59
requirements. So for example, last year there was
6:01
a major blood culture shortage
6:04
nationwide. So we're able to use DGN support
6:07
to help reduce
6:08
the blood culture bottle usage very significantly.
6:12
And now we are spending to do more
6:14
for daily labs and any other things to
6:16
reduce these unnecessary labs, which will be important
6:19
to solve to reduce the cost for the
6:22
hospital.
6:23
Also in the meantime,
6:24
it's also reducing unnecessary blood draws or reduced
6:27
ultraglionic anemia and obviously patient experience
6:30
will be better without these additional plastics as
6:33
well.
6:34
So these are things I would just you
6:36
know, maybe call out particularly, but there's really
6:38
a lot going on,
6:40
right now.
6:43
That's great to hear. And, you know, really
6:44
cool to,
6:46
have the ability to leverage technology in ways
6:48
that have a meaningful impact on patient care
6:51
as well as the workflows on the clinical
6:52
side. I I know reducing the burden on,
6:55
clinicians, doctors, and nurses especially,
6:59
is truly a mission for so many organizations
7:01
overall.
7:02
So that those are both great examples to
7:04
hear about. I'm curious, looking ahead over the
7:07
next twelve months or so, where do you
7:08
see some of the big biggest growth opportunities
7:10
for you and your teams?
7:13
Yeah. So, with the joint partnership with,
7:19
IT, marketing and informatics and data analytics,
7:22
we formed what we call the Center for
7:24
Digital Transformation.
7:26
This is a center
7:28
we established
7:30
and to really help us
7:31
adapt to the a lot of the new
7:33
trend
7:36
in the healthcare right now. So obviously lots
7:38
of work about digital transformation about how we
7:40
actually deploy this digital
7:42
health and AI technology
7:44
in the clinical care and really improve both
7:47
the provider patient experience
7:49
and improve the efficiency of the health system.
7:52
So there's been a lot of initiatives sort
7:54
of under the umbrella for this,
7:56
what we call the CDP,
7:58
Center for Digital Transformation.
8:00
And we're also looking at how do we
8:02
review the roadmap
8:04
of the digital tools we are thinking about
8:06
that can help bridge the gap and the
8:08
pinpoint that we have in the system. We
8:10
are looking at establishing
8:12
AI governance for the increasing
8:15
number of AI usage, right, almost in every
8:17
industry. We're also looking at how we do
8:20
a better job in terms of intaking some
8:21
of these solutions.
8:23
There's always
8:24
lots of DG Health startups out there and
8:26
they probably download a number in the last
8:28
few years. So how do we making sure
8:30
we have a process where we can
8:34
broadly taking those in and in the meantime
8:36
be able to evaluate them fairly and
8:39
really factor into
8:41
decisions about what's the best
8:44
really to help solve the problems that we
8:46
have in our health systems, what's the best
8:48
model to work with the vendor to co
8:50
develop
8:51
some of these products, because there there's a
8:53
lot of new technology and none of them
8:55
are mature yet. In the meantime, we also
8:57
have
8:59
some homegrown resources that we can build things
9:01
as well. So thinking about how we build,
9:04
when do we build, when do we buy
9:06
is also the other thing we want to
9:07
sort of factor into this new process of
9:11
that innovation intake.
9:12
So that's I think that's a really exciting
9:14
opportunity for our team to think about how
9:17
we can help institution prioritize these opportunities
9:20
and really help bridge the gap and make
9:22
a true impact, on clinical care.
9:27
I love that. I I think it's so
9:29
critical, as you mentioned, to have, you know,
9:31
the right evaluation process to bring in technologies
9:33
because there are so much exciting things out
9:35
there and new
9:36
possibilities, especially with AI,
9:39
and automation. But, you know, not everything is
9:42
gonna be beneficial. And so being able to
9:44
stay on track and really understand,
9:47
what makes the most sense
9:48
is so important.
9:49
I'm curious. You mentioned, you know, getting some
9:51
of the the governance processes in place,
9:54
especially for AI use. Could you talk a
9:57
little bit more about that and what those
9:58
conversations have been like at UChicago Medicine?
10:01
Yeah. So that's a great question regarding the
10:03
AI governance. So maybe to start with, you
10:06
know, thinking about why we need that. So,
10:08
right, there's a lot of AI companies out
10:10
there. And almost like when I go to
10:12
a conference, it's like almost every company have
10:15
AI tagged in their sort of product description.
10:19
It's almost as if this is something that's
10:22
universally used right now.
10:24
But
10:25
when we look at these AI solutions, first
10:27
is, is it really addressing a critical problem
10:30
that we have? And so secondly,
10:32
is that problem really need to be solved
10:34
by AI? So that will be an important
10:36
piece that we need to tease out when
10:38
we think about the intake, think about
10:41
evaluating these solutions. But after we sort of
10:43
decide, all right, there is a true pain
10:45
point and problem we're trying to solve And
10:47
we do need a solution in a place.
10:49
And AI seem to be a potentially a
10:50
good
10:51
solution to solve the problem. We still would
10:54
like to take a very thoughtful process in
10:56
terms of validating is AI really delivering what
10:59
it's supposed to do. So accuracy, completeness, right,
11:02
all these type of evaluation about
11:05
the effectiveness of AI. There's also a lot
11:07
of risk assessments,
11:10
especially nowadays with a lot of cyber
11:13
security risk.
11:15
So making sure we are reviewing this AI
11:17
in a fair way to adjust assist their
11:20
to assess their risk level
11:22
based on whether the use of AI is
11:24
susceptible, not susceptible and high risk versus low
11:27
risk.
11:28
And also in the meantime making sure they
11:30
meet our security, privacy and
11:32
asset standards are sort of an important piece
11:35
in the intake of process as well. And
11:37
last but not least, let's say something really
11:39
works well
11:40
in past all the risk assessments that we
11:42
do, we still want to have a process
11:44
to continue to monitor this AI moving forward
11:47
to making sure they still perform
11:50
as the same way we thought it would
11:51
be in the beginning.
11:53
Health system is constantly changing. So all these
11:55
changes regardless of spending the new units, building
11:58
a new hospital,
12:00
adding additional outpatient site, we all potentially can
12:04
change
12:06
how AI model works. We want to be
12:08
able to have some light of sight into
12:09
that too. So these are the things we
12:11
thought about having a strong AI governance. And
12:14
again,
12:15
the people who will be overseeing the AI
12:17
governance will be multidisciplinary.
12:19
So know, people with, you know, data science,
12:22
you know, analytics backgrounds, informatics, ITs,
12:25
and, you know, on security, privacy, and other
12:27
stakeholders will also be involved as well.
12:32
Got it. That's helpful to to know. Thank
12:34
you so much for digging a little bit
12:35
deeper in there. Now it really seems like
12:37
a lot of opportunities out there, a lot
12:39
of cool things that are happening, but I
12:41
can imagine there are challenges as well. What
12:44
are some of those challenges or roadblocks that
12:46
you're anticipating
12:47
in the
12:48
coming year or so?
12:50
Yeah. We are taking a very thoughtful way
12:53
when we deploy this technology.
12:55
So,
12:56
you know, first, like we talk about, the
12:58
patient community that we serve. Right? It's not
13:01
exactly the most digitally,
13:03
savvy population.
13:05
So making sure every time we roll out
13:06
something,
13:07
we'll
13:08
always have the digital equity
13:12
things in mind that we want to make
13:13
sure the things we deploy are benefiting
13:16
all the patient population,
13:18
maybe not all through the same venue, but
13:20
always making sure there's a way to engage,
13:22
making sure there's a way to help them
13:24
navigate
13:26
and benefit from all the new solution we
13:28
deploy throughout the health systems. So I think
13:30
that
13:32
all the work around digital equity
13:35
and try to
13:37
overcome all the challenges
13:39
from digital device, something I think we have
13:41
seen since the pandemic and also something we're
13:44
still compelled to continue to improve
13:47
through our processes.
13:49
And now with AI, right, there's even more
13:52
thing to think about around
13:54
the around like the biases,
13:58
based on different gender, race, populations,
14:02
etcetera, like how does AI use
14:06
being properly governed
14:07
to making sure that we are free of
14:09
these biases
14:10
or ethical concerns? It will be the other
14:12
challenge we need to constantly test out and
14:15
just be mindful of in our processes.
14:18
So I would say I wouldn't say that's
14:19
necessarily a challenge. Instead, I think these are
14:22
all good questions that we have been solving
14:24
over the years before you even have this
14:26
technology.
14:28
But really with all the tools in mind,
14:30
how can we even use this tool in
14:32
the right way to actually mitigate some of
14:35
these gaps and make things better, not the
14:37
other way around? It's just something we need
14:39
to constantly be careful about and margin when
14:41
we march forward.
14:46
Got it. That's helpful to know. And and
14:48
certainly,
14:49
as you can imagine oh, I know a
14:50
lot of goes into trying to figure out
14:52
those AI models and, how you can do
14:55
that in an equitable way,
14:57
especially given how quickly the technology is evolving
14:59
and changing.
15:02
Yeah. Exactly. I mean, it's it's,
15:05
rapid moving world right now. So,
15:08
it's, again, both
15:10
a challenge, but in the meantime, any excitement.
15:12
I feel actually very fortunate to,
15:15
live in this era where it's witnessed, like,
15:17
all the Internet growth and now AI growth,
15:21
and it's just gonna change a lot of
15:22
things, which is super exciting.
15:26
Absolutely. That's fantastic to hear. It is a
15:28
great time to be in health care for
15:30
sure. Now I'm curious from your perspective, given
15:33
some of the, big,
15:35
things that we're excited about as well as,
15:37
you know, challenges that we just mentioned,
15:39
What is the number one thing that you're
15:41
doing right now to set UChicago Medicine up
15:43
for long term success?
15:45
Yeah. I will talk about one project I'm
15:47
working on that's actually related to what we
15:49
talk about. So we talk a lot about
15:52
the use of AI is growing and there's
15:54
more companies, there's more use cases here and
15:57
there that's coming up as we speak. And
15:59
we also talk about I mean, there's a
16:01
lot of saying around people with the AI
16:04
may replace
16:05
the people who don't know how to use
16:06
AI. And maybe in the future, there will
16:08
be, you know, even more, AI and automation,
16:11
right, in in the work. So I think,
16:13
absolutely, what we want to do is making
16:15
sure we elevate the AI literacy in our
16:17
institution. Right? We want to train people to
16:19
know how to make the best use of
16:20
AI.
16:21
AI is not perfect. So training them to
16:24
understand
16:26
what's the posting counts, what's the limitation, what
16:29
does it mean by AI hallucinates.
16:32
And but in the meantime with all of
16:33
that, being able to use leveraging AI as
16:36
most most likely your inexperienced intern. Someone is
16:39
on your team. It's almost like your digital
16:42
worker, but who is inexperienced, who is new
16:44
to the team. So how do you train
16:45
the AI to actually work with you to
16:47
actually get things done more efficiently?
16:49
That that will be sort of the goal.
16:51
We we hoping that we,
16:53
convey to our sort of colleagues. So one
16:56
thing I'm working on is actually building
16:58
up
16:59
generative AI platform that will be HIPAA compliant
17:01
and secure
17:03
for internal use, which will allow us, meaning
17:05
our staff
17:07
and health care provider being able to upload
17:09
certain
17:11
information about PHEIs
17:12
and be able to use that to
17:15
do things, for example, like
17:18
drafting some of the messages that we are
17:20
about to send out in different formats, in
17:22
different languages, in different
17:24
type of literacy level, which allowed them to
17:27
help improve some of this patient communication.
17:30
Or for example, we are developing some of
17:32
the modules to help drafting denial letters of
17:34
appeals
17:35
or for example answering some questions based on
17:38
our policies and procedures.
17:40
So while we are standing that platform up
17:42
hopefully in the next
17:44
few months, it's really coming up soon,
17:46
we also want to pair that with
17:48
staff communication,
17:49
education about how you should use AI, right?
17:52
How to prompt,
17:53
what's the AI governance we've been talking about
17:55
so far? And when you have questions, where
17:58
do you reach out? And how do you
18:00
use the tool better to your
18:03
for your purposes to to boost your efficiency
18:06
in every area. So that's something I think
18:08
really exciting, and I think I'm gonna spend
18:10
quite some time really to help stand it
18:11
up and thinking about creating different use cases
18:14
and,
18:15
and training people to to make use of
18:17
it. But what we do, I think it's
18:19
gonna be really,
18:20
improving a lot of efficiency issues that we
18:22
have and really allowing people to really do
18:25
things that are probably a license. Right? It's
18:27
like you let AI to do a lot
18:29
of these repetitive
18:30
or something that's, you know, less,
18:34
less challenging
18:35
per se in terms of the complexity or
18:37
originality.
18:38
But, you know, reserving all your energy on
18:40
something that really requires all these, you know,
18:43
things,
18:44
and the AI just help you out to
18:45
save time. So, that's something I think really
18:48
exciting, that I'm really working on right now
18:50
that I think really would set our institution,
18:53
at a really good start
18:55
as we continue to, spend and,
18:58
deploy additional AI solutions.
19:02
That makes a lot of sense. You know,
19:04
it's really, helpful to understand from a leadership
19:07
context and perspective,
19:08
how you're thinking about these things and really,
19:10
really putting yourself in a strong position to
19:13
continue to,
19:15
advance this technology and bring it forward in
19:17
a meaningful way. Kai, thank you so much
19:19
for joining us on the podcast today. This
19:20
has been such a fun conversation, and I
19:22
look forward to connecting with you again soon.
19:24
Thank you so much, Laura. And likewise here,
19:27
meeting you, and, have a good day.