Latest / Becker’s Healthcare Digital Health + Health IT / Ratnakar Lavu, Chief Digital Information Officer at Elevance Health

Transcript

Ratnakar Lavu, Chief Digital Information Officer at Elevance Health · Becker’s Healthcare Digital Health + Health IT

0:00
Today's episode is brought to you by Abridge.
0:04
Named best in class for the ambient AI
0:06
segment, Abridge is the leading AI platform for
0:09
clinical conversations
0:11
designed by clinicians and scientists
0:13
to meet the diverse needs of the largest
0:16
and most complex health systems.
0:19
Abridge automates clinical documentation
0:21
in real time and integrates directly within the
0:24
electronic health record workflow,
0:26
reducing the burden and distraction of repetitive administrative
0:30
tasks across care settings, specialties,
0:33
and languages.
0:34
A bridge has been shown to reduce burnout
0:36
by 55%,
0:38
improve clinician job satisfaction by 85%,
0:42
and significantly
0:43
increase patient satisfaction scores according to surveys from
0:47
Press Ganey.
0:48
Bring this impact to your health system today
0:50
and join the community of users across leading
0:52
health systems, including Duke Health, Emory Health,
0:56
John Hopkins Medicine, and Mayo Clinic. Visit abridge.com
1:00
to learn more today. Hello, everyone, and welcome
1:03
to the Becker's Healthcare Podcast. I'm Scott King
1:05
joined by a very, very special guest today,
1:09
Ratnakar Lavu, chief digital information officer with Elevent's
1:13
Health. Ratnakar, thanks so much for joining us.
1:15
Really excited to have you. How are you
1:16
doing?
1:17
Good, Scott. Thank you for having me here
1:20
today.
1:20
Really excited to talk through things with you.
1:24
Yeah. We have we have some big things
1:25
to get into with, you know, on in
1:27
the payer
1:28
forefront and in health care and technology. But
1:31
before we dive in there, can you please
1:32
just tell us a little bit about yourself
1:34
and your background?
1:36
Yeah. So,
1:37
I've
1:38
my name is Raka Karlavu. I'm the CDIO,
1:41
here at Alvan's Health, and I've always been
1:43
within technology and really focused on how we
1:47
can leverage technology, data, and AI,
1:51
to simplify actually,
1:53
consumer experiences and member experiences.
1:56
And here,
1:57
I'm really focused on simplifying
1:59
health care to improve outcomes, leveraging technology, data,
2:03
and AI. We're embedding AI responsibly
2:06
across all our operations,
2:09
not just as an experiment, but we're doing
2:11
it at scale
2:12
to make care more affordable,
2:14
accessible,
2:15
and personalized, Scott. So that's what's actually really
2:18
exciting about the work that, we do here
2:20
at Eleventh Health.
2:22
Yeah. Yeah. I I feel like in health
2:24
care, we're hearing just more and more about
2:26
how vital, you know, especially on a payer
2:27
side, data is and then across the board
2:30
in the industry,
2:32
AI and the the uses there, and and
2:34
we're just tapping into that. But what do
2:36
you have
2:37
your eye on right now as far as
2:39
opportunities go?
2:41
Well, Scott, you know, health care is very
2:44
complex. Right?
2:45
And, every one of us, who interact with
2:48
health care,
2:49
see some of this complexity, whether it is
2:52
trying to understand the benefit,
2:54
trying to connect with the provider
2:56
to get care,
2:58
and really focused on kind of, you know,
3:00
the health care outcomes itself or the clinical
3:03
outcome. So in, Eleventh Health, we're really focused
3:06
on three things.
3:08
One is simplifying
3:10
kind of the member experiences itself. We wanna
3:13
create personalized and more seamless experiences.
3:16
We wanna actually simplify the,
3:19
care provider experiences also, the interoperability.
3:22
We wanna empower care providers to improve health
3:25
outcomes.
3:26
And then finally, we wanna simplify
3:28
how we work to actually better serve our
3:31
customers and members. And I'll give you a
3:34
couple of examples.
3:35
Let me start with,
3:37
the member or the customer,
3:40
example itself.
3:41
So we have something called,
3:44
proactive member engagement.
3:46
And,
3:47
what this is is it's actually an AI
3:50
driven tool
3:51
wherein we identify members who could benefit from
3:55
an outreach
3:56
to proactively address care gaps. So we study
3:59
the members, their clinical data. We understand
4:03
who could actually benefit from closing care gap.
4:06
We do an outreach for to them
4:08
to explain to them what would be beneficial,
4:13
as they think about going to their care
4:15
provider,
4:16
what will be beneficial for them to address
4:19
so that they can actually have that meaningful
4:21
interaction with their care provider,
4:24
with a lot more information. And we do
4:26
millions of these outreaches
4:27
today.
4:28
And it's the AI, the in data,
4:32
and then the kind of the intelligence that
4:34
we build that enables that to happen. The
4:36
other example is when members actually call in
4:39
to our call center,
4:41
we provide our call center agents with the
4:44
intelligence
4:44
to be able to actually,
4:47
address the member issues. So we we correlate
4:50
all the information,
4:51
the clinical data,
4:53
you know, kind of their,
4:55
claims data,
4:56
their benefits data, and automate the things,
5:00
for our, call center agent.
5:03
Not only that, but once they finish the
5:05
call, we actually do a post call wrap
5:08
up,
5:09
which is we use AI to understand how
5:12
the call went.
5:13
How did our members feel?
5:15
What were their issues?
5:17
And we do millions of these, post call
5:19
wrap ups,
5:20
every month. And the reason for that is
5:23
we actually understand the sentiment of the member
5:26
so that we can proactively
5:27
address that.
5:29
The reason these examples are important is because
5:31
we're very focused and obsessed about the member
5:34
experience, the member interactions itself. And I'll then
5:38
talk a little bit about the care providers.
5:40
So in the care providers, we wanna make
5:43
the interaction between us and the care providers
5:46
easy. So within
5:48
UMAI, where they actually,
5:51
kind of submit their prior on, we actually
5:54
look at what they've submitted,
5:56
analyze it with the data that we have,
5:59
and our entire goal and objective
6:01
is to approve things as fast as we
6:04
can, for the care provider.
6:06
And,
6:07
we use a lot of AI to look
6:09
at medical record, to look at their prior
6:11
auth submission,
6:12
to look at the member's benefits data, and
6:15
orchestrate that approval
6:17
in a more seamless way for the care
6:19
provider.
6:20
And another thing is the care providers actually
6:23
submit a lot of claims. Our providers submit
6:26
a lot of claim. Sometimes those claims are
6:29
not complete.
6:30
And so we have at applied AI
6:32
to look at those claims and see where,
6:35
the information is not complete. And the AI
6:38
actually help in
6:40
completing that information so we can process the
6:42
claims on behalf of the providers. Why is
6:45
this important? Because a lot of care providers
6:47
spend a lot of time in administrative work,
6:50
and we wanna take away that or simplify
6:52
that administrative work for the for them so
6:55
that they can really take care of our
6:56
members and,
6:58
close those care gaps that are extremely important
7:01
for our members.
7:02
Now let me come to how we are
7:05
simplifying
7:06
the work itself for our associates.
7:08
So we have we're deploying a lot of
7:11
AI productivity tools so that they we can
7:14
automate a lot of their workflows
7:16
so they really can actually service our members
7:19
and providers.
7:20
So we have over 52,000
7:22
associates today.
7:23
We will that are leveraging some sort of
7:26
AI capabilities,
7:28
within their workflows. But we're also rolling out
7:31
now,
7:32
chat GPT across the enterprise.
7:35
So not only is it for that that
7:37
they can do their work more easily,
7:40
but they can also kind of understand leveraging
7:43
in these chat GPT and other capabilities
7:46
to be able to serve our members effectively.
7:48
And then the not only are we rolling
7:51
out these tools,
7:52
we're actually going to take them through AI
7:54
certification, working with OpenAI.
7:57
Because one of the things we find is
7:59
it's not just about the tool, it's our
8:02
associates need to understand how to use those
8:05
tools, what what it means to leverage AI
8:08
to actually do their work better, and that's
8:11
where we're we're focused on quite a bit,
8:14
with our associate.
8:15
So, again, to recap,
8:18
we want to actually create a more personalized
8:21
and more seamless experience for our members. We
8:24
wanna empower our care providers to improve health
8:27
outcomes,
8:28
and then we wanna simplify
8:30
how we work to better serve our
8:33
customers and members itself.
8:35
Incredibly fascinating. A lot of so much great
8:38
information in there right in the car. Let
8:40
let me ask you, I guess, to play
8:41
and to play devil's advocate a little bit,
8:43
you know, even the most practical uses of
8:45
AI the average person uses, there's still some
8:48
some trial and error. So just curious with
8:50
something like the the, AI member outreach that
8:53
the elephants has been doing and even those
8:55
summaries that it provides, what what's kind of
8:58
some of the, I guess, the errors that
9:00
there might be, and are they kind of
9:03
fixed through the AI self learning? How's how's
9:06
that working? How's it getting kinda past any
9:08
any hiccups there?
9:09
Yeah. Actually, I I wouldn't say it's a
9:11
lot, about the hiccups. What we have learned
9:14
actually is we need to obsess about the
9:16
journeys themselves.
9:18
So we are really, kind of focused on
9:20
the member journeys,
9:22
overall or the provider journey. So I'll give
9:24
you an example. In the UMAI
9:26
case,
9:27
when, a prior auth is submitted,
9:30
our entire focus is to how to get
9:33
to real time prior authorization. Right? So when
9:36
we focus on that journey,
9:38
what we have found is when we deploy
9:41
AI solutions,
9:42
we our our own learning has been
9:45
don't just deploy it in in a part
9:48
of the workflow,
9:49
really think about the entire workflow. So when
9:52
we deployed it in the case of prior
9:54
auth,
9:54
we were only looking at it as the
9:57
kind of the
9:58
approval for the prior auth. In in fact,
10:01
there's a lot more in the process that
10:03
we'll have to think through. And so we
10:05
went and pivoted.
10:07
Instead of just a portion of the workflow,
10:10
we pivoted to actually looking at the entire
10:12
workflow.
10:13
And that has been the big learning for
10:16
us, is when you think about AI, just
10:18
don't think about it in a silo of
10:20
a workflow.
10:21
Think about the entire workflow,
10:23
and
10:24
we
10:25
reimagine that workflow
10:28
as you think about,
10:29
building solutions and deploying AI solution. And that's
10:32
how actually you'll make the entire workflow simple.
10:36
We we we're re and then we'll also
10:38
get them to be more,
10:40
kind of connected
10:42
as we think about the interoperability
10:44
between provider and payer.
10:46
Right. It's a journey. So focus on that
10:48
journey. I I I think that makes complete
10:50
sense.
10:51
In in terms of, you know, the potential
10:54
maybe that people or payers have even tapped
10:57
into
10:58
for AI, what are some of the things
10:59
that Evolence is kind of looking to in
11:02
terms of focus for the future?
11:05
Well, I see tremendous amount of potential,
11:07
in terms of the focus for the future.
11:09
Again, we're deep rooted into creating personalized and
11:13
more seamless experiences.
11:14
So I'll give you an example. We we
11:17
have deployed a virtual
11:19
assistant, in our Sydney app
11:21
for our members to use that. It's like
11:23
a chat GPT experience. So they can actually
11:26
ask questions. Is is this particular procedure covered
11:29
in my benefits?
11:31
Right? And then we we kind of the
11:33
AI has intelligence to bring that information back
11:36
to the member.
11:38
And then not only that, it act it
11:40
can then connect that member
11:43
to the right care provider
11:45
so who can drive quality outcomes at a
11:47
at an affordable price. So we have built
11:49
that intelligence where the AI understands what the
11:52
member is asking for and how to connect
11:54
them to get the care and
11:57
the clinical outcomes
11:59
that,
12:00
we we actually want to guide the member
12:02
through. And so what I'm excited about is
12:05
this creating this personalized and more seamless experiences
12:09
at scale
12:10
now leveraging AI. The other thing that I'm
12:12
really kind of,
12:14
excited about, as I mentioned,
12:16
is improving this interoperability
12:19
between here and provider because we wanna streamline
12:22
and make it simple
12:24
for the,
12:25
providers to actually engage with us. And I
12:28
think there's tremendous amount of potential there. And
12:31
finally,
12:32
I'm really,
12:33
kind of excited about how we can transform,
12:37
internally our company
12:38
so that we can kind of automate a
12:42
lot of the mundane tasks,
12:44
that are there, within a large company like
12:46
ours with, leveraging AI, and really focus on
12:50
on the member and the provider and the
12:52
high value added task because that's where we
12:55
need to be spending a lot more time
12:56
to simplify health care.
12:59
Yeah. And for anyone listening who wants to
13:00
to get more information or or read about
13:02
this virtual assistant Ovens is working on, Retina
13:05
Carr did a great interview with Becker's, looks
13:07
like the end of of October. And,
13:09
let let me ask you. How's how's it
13:11
how are the
13:13
the total number of members that'll have access
13:16
to the virtual assistant? How's that tracking as
13:18
we're here almost, you know, at the 2025?
13:21
Actually, it's tracking really well. So we're getting
13:24
a lot more adoption,
13:26
you know, as we roll this out to,
13:29
kind of,
13:30
a large set of our members, we're getting
13:32
a lot of adoption,
13:34
for the virtual assistant.
13:36
What we are seeing is the members are
13:38
highly engaged. They wanna understand their benefit. And
13:41
the first for the first time, we made
13:43
it actually extremely simple for them to understand
13:45
the benefit.
13:46
Not only that, it is actually making them
13:49
easier also to connect with the right provider
13:52
because the intelligence that we have built through
13:54
this
13:55
actually connects them with the right provider who
13:57
can drive high quality of care at an
14:00
affordable price. And so we're seeing a lot
14:02
of engagement. What actually it's helping us do
14:06
is the member is able to,
14:08
kind of do self-service
14:10
instead of calling and trying to understand their
14:13
benefits or calling to actually get to the
14:16
right provider. Now all of it is moving
14:18
towards self-service, and that's what we're really excited
14:21
about. Because we really wanna empower our members,
14:25
to kind of navigate this complex health care
14:27
system, and we wanna make it easy and
14:29
simple for them to do that.
14:31
Yeah. It sounds like the rollout has been
14:33
going great.
14:35
As organizations
14:36
continue to expand and implement AI driven solutions,
14:39
how do they do so in a responsible
14:41
and ethical way?
14:44
It's a great question, Scott. So, by the
14:45
way, we begin with
14:47
responsible AI. We begin with transparency.
14:51
We begin with explainability.
14:53
So we actually have
14:55
an entire responsible AI
14:57
governance framework within our company. So every AI
15:01
solution has to go through that governance
15:04
framework where we evaluate,
15:07
the AI solution
15:08
for,
15:09
responsible
15:10
AI, transparency, and explainability,
15:13
and how the solution is actually designed and
15:16
being implemented.
15:17
Only then will we actually approve that AI
15:20
solution
15:21
to go into production.
15:23
And then we monitor that,
15:25
you know, once it's in production
15:26
to ensure that all the transparency and explainability
15:31
is are
15:32
has been implemented
15:34
effectively once it goes into production. So you
15:36
need to both have a framework in terms
15:39
of the governance framework.
15:41
That governance framework has to start from the
15:43
beginning of the solution when it's actually designed,
15:46
not after the fact. And so it has
15:49
to be part of the process.
15:50
And then once you actually implement the solution,
15:54
you actually have to also monitor it for
15:56
its effectiveness.
15:57
And that's the way we have actually done
15:59
that, and we've been very successful
16:02
in doing that,
16:03
across all our AI solutions.
16:06
In terms of all the uses Elvents has
16:08
had for AI, a lot of the things
16:10
you've discussed already,
16:12
has Elvents been able to pinpoint,
16:15
like, you know, just a a number for
16:17
efficiency in regards to hours saved between
16:21
payers and and and members?
16:23
Yes. We do. And so one of the
16:25
things we constantly watch is,
16:28
is it improving
16:30
the outcomes that we we have kind of,
16:34
outlined,
16:35
you know, in terms of the member efficiency
16:38
to actually self serve. Right?
16:41
That's a huge efficiency because the members,
16:44
are effectively
16:45
getting all the information that they had to
16:47
go to multiple places in the past. So
16:49
self-service.
16:50
And then self-service, not only just self-service, how
16:53
satisfied are they about with the self-service, so
16:56
NPS scores and other thing. So we monitor
16:58
all of that. We also look at in
17:00
the peers,
17:03
in the peer provider interaction,
17:05
we also looked at kind of the effectiveness
17:08
of our interaction overall. How,
17:10
satisfied
17:11
are the providers,
17:13
both in terms of how we
17:15
kind of set them up to begin with,
17:18
and then how do they,
17:19
actually how do we process
17:21
their kind of, prior auths, claims,
17:25
and and any other administrative thing. We actually
17:28
look at how effective those are, how are
17:31
we reducing the time of processing those,
17:34
administrative
17:35
tasks, and then how satisfied are the providers
17:38
there. And then in the associate population, we
17:40
do a lot of,
17:43
self assessment.
17:44
How satisfied are are
17:46
our our associates
17:48
in actually
17:49
looking at, the tools that we've deployed, and
17:52
is it actually helping them to do their
17:54
job better? So you we monitor a lot
17:58
of KPIs
17:59
across
18:00
the entire ecosystem.
18:01
And, again, I come back to
18:04
these
18:05
these the solutions that we're deploying,
18:08
the KPIs that we're monitoring
18:10
actually yield to a better and more simplified
18:13
experience
18:14
for all. And and and I think that's
18:15
a way to actually simplify health care,
18:18
for everyone.
18:20
Yeah. It really seems that for all these
18:22
uses, you're you're mentioning that Evans goes to
18:25
great lengths to get member feedback. How crucial
18:28
is that feedback kinda during the early stages
18:30
of some of these rollouts?
18:32
It's actually Scott, it's
18:35
I'm glad you brought that point up because
18:37
it's actually extremely important
18:40
to get this feedback, but both from members
18:42
and providers.
18:43
And we
18:45
we actually study so even before rolling out
18:48
a solution like the virtual assistant,
18:50
we've done a lot of market analysis. So
18:52
we brought in members to say, will this
18:55
even be
18:56
valuable to you?
18:58
Is this going to simplify
19:00
in how you kind of look at your
19:02
benefits, how you can connect your provider?
19:04
So we do a lot of market research
19:07
and assessment.
19:08
And then when we deploy it, we get
19:10
a lot of this feedback,
19:12
and our teams are really obsessed about member
19:15
feedback and provider feedback.
19:16
Because we wanna make sure
19:18
that we are constantly listening to that feedback
19:22
to simplify these experiences.
19:24
Because that's the way we actually learn, and
19:27
we continue to evolve the solution
19:29
because we're in service of them. And and
19:31
we wanna make that, very easy
19:34
and simple
19:35
for our members and providers. And so this
19:37
feedbacks,
19:38
loop is actually very important for us in
19:41
how we evolve our solutions.
19:43
Last question I have for you, Radhnikar.
19:46
What do you think 2026
19:48
will bring for digital technology and AI?
19:51
Yeah.
19:52
I I think,
19:54
you know, as AI is evolving,
19:57
as we are evolving
19:58
kind of our thinking about data,
20:01
our understanding,
20:04
better about our members and providers and how
20:06
they interact with the tools and capabilities
20:09
that we're deploying,
20:10
we see tremendous amount of potential
20:13
to actually
20:15
streamline
20:15
and create
20:17
more personal,
20:18
a more connected and seamless experiences for our
20:21
members.
20:22
We also,
20:24
kind of are bullish on
20:26
simplifying the interoperability
20:28
with our providers so that we can empower
20:30
care providers
20:31
to improve health outcomes.
20:33
And finally, we really want to accelerate
20:37
a lot of the tool adoption
20:39
internally
20:39
and the capabilities internally
20:41
so that we can simplify our workflows
20:44
so that we can better serve our members
20:46
and providers. And I I'm actually very excited
20:49
about that because we've laid the foundation, and
20:51
now we believe we can accelerate
20:53
this transformation.
20:56
A 100%. It's so great to hear how
20:58
Elevent is kind of, you know, stepping on
21:00
a gas in terms of AI and leveraging
21:02
it for the member experience and then also
21:05
constantly gauging that member feedback, which is crucial.
21:08
The Randekar, this is a great conversation. You
21:10
know, we're ending now. I feel like I
21:11
could've talked to you for two hours about
21:12
this. So thanks so much, and really looking
21:14
forward to working with you again soon. Thank
21:16
you so much, Scott. It it was really
21:18
nice. Really excited about, our journey, and I'm
21:21
glad, to share that with everyone. Thanks a
21:23
lot.