Latest / Becker’s Payer Issues Podcast / AI-Driven, Personalized Care for Chronic Conditions and How Payers Are Adapting Through Digital Health Innovation

Transcript

AI-Driven, Personalized Care for Chronic Conditions and How Payers Are Adapting Through Digital Health Innovation · Becker’s Payer Issues Podcast

0:00
Hi, everyone. This is Lucas Voss with Becker's
0:02
Healthcare. Thanks so much for tuning in to
0:04
the Becker's Healthcare podcast series. It's fantastic to
0:07
have you. Today, we're talking about AI driven
0:10
personalized care for chronic conditions
0:13
and how payers are adapting through digital health
0:16
innovation.
0:18
And joining me for today's discussion, I'm so
0:20
excited to have her, is doctor Lisa Shaw.
0:22
She's the executive vice president and chief medical
0:24
officer of Twin Health. Doctor Shaw, thanks so
0:26
much for being here today. It's great to
0:28
have you. Thank you, Lucas. It's wonderful to
0:30
be here.
0:31
For those that might not know you yet
0:33
at least, I do wanna start off with
0:35
introductions.
0:37
If you could just introduce yourself and just
0:39
tell us a little bit about your work
0:40
in health care.
0:42
Yes. That's great. So, my name is doctor
0:44
Shah,
0:45
Lisa. I've been a I'm an internal medicine
0:48
and pediatrics trained physician.
0:51
I've been practicing for over twenty years now.
0:54
Primarily now internal medicine, and we'll talk a
0:56
little bit about what I do now. I
0:58
did my training at the University of Chicago,
1:00
and for the last, fifteen years, I've really
1:03
been focused
1:04
on designing and implementing
1:06
value based and and next generation models of
1:09
care, particularly around chronic conditions and and chronic
1:12
diseases.
1:13
The work I do at Twin Health is
1:15
very personal for me. We've been focused here
1:18
on chronic metabolic diseases, and we'll talk a
1:20
little bit about that today.
1:22
But both my parents have type two diabetes
1:24
and obesity,
1:26
and Twin has really been the first time,
1:28
you know, we we go into becoming a
1:29
physician to help the people we love the
1:31
most. It's been an incredible opportunity to really
1:34
see something
1:35
change their lives. They were the first two
1:38
patients on this program in the history of
1:40
the program. And so,
1:42
really doing great, reversed their diabetes, lost combined
1:45
45 pounds, and are off medication, and we'll
1:48
talk a little bit about that today. When
1:50
I started my career, I think one of
1:52
the things I really struggled with
1:54
was how impersonal medicine felt when I went
1:57
into practice.
1:58
Yeah. It you know, especially in primary care,
2:01
it felt very much as I was giving
2:03
generic advice
2:05
to people.
2:06
You know, eat better,
2:08
exercise more,
2:09
stress less, sleep more, and then they look
2:12
at me and go, thanks. Can I have
2:14
a prescription?
2:15
And that's really what was hard for me.
2:19
And if you fast forward to where we
2:21
are today,
2:23
the opportunity to use artificial intelligence
2:26
and machine learning,
2:27
techniques and sensors, and we'll talk about how
2:29
Twin does that today,
2:31
with compassionate
2:32
human care,
2:34
has really made this hyper personalization,
2:37
continuous real time care possible so that people
2:39
can actually,
2:41
you know, fix the disruption in their bodies,
2:44
fix this metabolic
2:45
damage,
2:46
not just manage a condition with a medication.
2:49
Yeah. And I do wanna talk about those
2:52
conditions a little bit more because I think
2:53
they're at the core of some of the
2:55
things that we're gonna touch on today in
2:57
our conversation. Again, we we talk about, as
2:59
you've mentioned, obesity and diabetes. Right? Unfortunately, those
3:02
numbers are continuing to rise. We have made
3:04
some inroads. We don't wanna ignore that, but
3:07
they continue to rise. Right? And when we're
3:09
looking from the pair lens too,
3:11
how are you seeing pairs respond there particularly
3:15
when it comes to some of the digital
3:17
innovation that you've just touched on? And then,
3:19
again, what's gaining traction for you,
3:22
and where are you seeing
3:24
gaps still exist?
3:27
Yeah. It's a great question. And, you know,
3:28
I've had been working with payers for most
3:31
of my career. And I think, you know,
3:33
in this value based care world, we started
3:35
with really traditional episodic care models and and
3:38
point solutions. Right? And that was a great
3:40
start if you look back fifteen,
3:42
twenty years ago,
3:44
a solution for diabetes, a solution for heart
3:46
disease, a solution for weight.
3:48
But I think there's a growing recognition that
3:50
those episodic and point solution,
3:53
models are really not bending the cost curve
3:55
enough, and they're not actually creating
3:57
a place for long term sustained
4:00
outcomes.
4:01
So payers are, I think, moving away from
4:03
that fragmented
4:05
point solution,
4:07
environment and landscape and really looking for two
4:09
things,
4:11
models that, are continuous and and wrap around
4:14
multiple conditions so that they're getting,
4:17
you know, a lot of value
4:19
for one,
4:20
solution
4:21
across all of these things that a that
4:24
a patient is struggling with. And then we
4:25
know these diseases do not travel alone. Obesity
4:28
doesn't travel separately from diabetes. They're partnered as
4:31
as is cardiac disease and some other diseases
4:34
we'll talk about today, hypertension.
4:36
And the second thing is I think they're
4:37
really looking for payment models that are standing
4:40
behind the outcomes and are outcomes based as
4:43
opposed to these traditional per member per month,
4:46
you know, type of models where you're really
4:48
paying, but there's no guarantee on the outcome.
4:52
So those two things have been really
4:54
what has,
4:55
shifted, I think, in the in the payer
4:57
landscape. I think the second thing is they're
5:00
seeing cost go up in the world around
5:02
things like medications,
5:04
and utilization
5:06
and prevalence. Right? So as you were saying,
5:09
despite our best efforts, diabetes and obesity are
5:11
growing. And
5:13
look at all the other conditions, metabolic associated
5:15
fatty liver disease growing rapidly, the number one
5:18
cause of liver transplants now in The US,
5:20
right, and then globally.
5:21
And so we're seeing these kinds of shifts,
5:24
and and they're scratching our heads. We all
5:25
are. So what do you what can you
5:27
really do to shift this new standard of
5:29
care or or create a new standard of
5:31
care? And I think,
5:33
metabolic health is now being seen as this
5:36
continuous,
5:37
suite of conditions that if we could really
5:40
figure out how to target
5:42
metabolism and metabolic health,
5:44
in a way that is very personalized because
5:47
it's unique for everyone,
5:48
there is an opportunity to really prevent,
5:51
both of the rising costs and the long
5:53
term health op, complications.
5:56
And I think in order to do that,
5:57
the traction we're seeing is really with solutions
6:00
that are data driven.
6:02
They're taking
6:04
multiple real time datasets
6:06
and personalizing
6:08
the engagement for the individual
6:10
to something that the person is willing to
6:12
do and can do,
6:13
that is also overseen by a clinical care
6:16
team, particularly for areas like diabetes and and
6:19
weight.
6:21
You mentioned something very critical there with with
6:23
just the member engagement piece, which is so
6:25
crucial because, again, it doesn't work if the
6:27
member doesn't engage. If there's no engagement, then
6:30
there's no solution really.
6:32
You talk about the concept of,
6:35
a digital
6:36
twin.
6:37
Can you
6:38
walk us through what that experience looked like,
6:40
what that is for members, and how it
6:43
differs from sort of the traditional
6:46
digital health approach that we're seeing a lot
6:49
around health care today?
6:51
Yes. Yeah.
6:52
It's a great question.
6:54
So digital twins have been around for a
6:56
long time. Twin, Twin Health did not create,
6:59
the first digital twin concept, though we are
7:02
the first,
7:03
application of digital twin technology
7:06
to human metabolism.
7:08
But digital twins are really they're similar technology
7:11
being used in other
7:17
self driving, autonomous driving cars,
7:20
have digital twins. And, essentially,
7:22
when you think of a very complex system
7:25
like
7:26
self driving cars or aviation or human metabolism,
7:30
arguably the most complicated system,
7:33
in the world.
7:34
The,
7:35
ability to use sensor technology, which we've really
7:38
evolved with here in health care right now
7:41
and in in
7:42
human body wearables.
7:44
But the the cars, for instance, put sensors
7:47
on the outside of the car and the
7:48
inside. And each driver is unique, but so
7:50
is the external environment where the car is.
7:53
Right? Whether you're in a traffic jam or
7:55
you're driving through snowy mountains or on a
7:57
beach.
7:58
So the ability to really predict using sensor
8:02
technology that gets smarter
8:04
with every drive and every,
8:06
exposure
8:07
to predict bad things before they happen is
8:10
really what a digital twin system is trying
8:12
to do, and then making
8:14
improvements before those bad things happen,
8:17
based on what it's seeing through all the
8:19
sensor data.
8:20
In metabolism,
8:21
we're doing the same thing. So we essentially
8:24
create an AI digital twin
8:27
of every Twin Health member
8:29
that evolves and learns every day
8:32
and really pays attention to clinical data that
8:35
the sensors are collecting.
8:37
So think about those as the inside of
8:39
the car or inside the human. Right? So
8:42
my blood pressure, my blood sugar,
8:44
my heart rate,
8:46
my stress levels, my sleep data that we're
8:49
getting from all of these wearables.
8:51
And then it's also looking at the outside.
8:53
So think of the environment where the car
8:55
is driving, the choices a person is making.
8:58
So am I choosing to eat at McDonald's?
9:00
Am I choosing
9:02
to,
9:03
sleep during the daytime because I have a
9:05
night shift job?
9:07
And so when it looks at all of
9:08
those things, whether controllable in the person's life
9:11
or not, but the reality of their life,
9:14
it puts the two together and creates models
9:17
that can help improve those clinical data points
9:20
with environmental,
9:22
lifestyle suggestions that the member is willing and
9:25
able to do. That's the key, willing
9:28
and able to do. Because if they're not,
9:30
we've seen this before. We've seen keto diets.
9:33
We've seen, you know,
9:35
extreme gym programs, and they work for a
9:38
few months and then people stop because they're
9:40
just not really sustainable.
9:43
I like to think that my extreme
9:45
gym program would work for a month. It
9:46
works for about a week.
9:48
So I completely understand. It's very, very important.
9:50
I agree. Again, we come back to the
9:52
engagement piece. Now the great thing is you've
9:54
touched on the fact that data is so,
9:55
so critical.
9:57
And
9:58
you announced results from a randomized controlled trial
10:01
with Cleveland Clinic, and it was published in
10:03
the New England Journal of Medicine Catalyst.
10:05
I'd love to talk a little bit about
10:06
that because I think that data is really
10:08
crucial to what we're talking about today. At
10:10
a high level, can you outline some of
10:13
those key findings
10:14
and sort of how they shape the broader
10:17
conversation about some of the things that you've
10:19
touched on specifically when we think about AI
10:21
driven personalized care?
10:23
Yeah. Yes. Absolutely. So the Cleveland Clinic study
10:26
that we did was very intentional.
10:28
We really wanted to demonstrate you know, we
10:30
see a lot of randomized controlled trials looking
10:32
at medications.
10:34
We don't see a lot of randomized controlled
10:35
trials where we can actually use,
10:38
AI digital twin technology,
10:41
lifestyle
10:41
changes,
10:43
and actually eliminate medications, which is really what
10:46
twin does with this AI digital twin model.
10:48
It gets people off medications.
10:50
It reduces their a one c.
10:52
It reduces their weight. They lose weight in
10:55
a healthy way, which we'll talk a little
10:57
bit more about,
10:58
later.
10:59
But it also is, healing a lot of
11:02
different aspects of their metabolism and their pathways
11:05
in their body.
11:06
So in this study, we wanted to demonstrate
11:08
that versus usual care for people living with
11:11
type two diabetes,
11:13
the AI digital twin intervention was superior.
11:16
And, we randomized people to either usual care
11:19
at the Cleveland Clinic
11:21
or the AI digital twin. Cleveland Clinic ran
11:24
the study.
11:25
The investigators did all of the randomization and
11:27
consenting.
11:29
100 people were randomized into the intervention arm.
11:32
And what we found is, that seventy one
11:35
percent of the people who were in the
11:36
intervention
11:37
arm normalized their a one c and got
11:39
off all of their medications
11:41
except metformin,
11:43
which is sort of a standard baseline,
11:45
compared to only two percent in the control
11:47
group. So seventy one percent of the intervention
11:50
arm at one year had an a one
11:52
c that reduced below 6.5, which is a
11:54
non type two diabetes a one c and
11:57
were off their high cost medication,
11:59
all medications actually other than the metformin versus
12:02
two percent in the control group. We also
12:04
found sixty six percent of participants
12:07
normalized their hypertension.
12:10
Eighty five percent of them eliminated the GLP
12:13
one medications
12:14
and sustained
12:15
weight loss. Not only that, but they saw
12:18
27 pounds average weight loss that they kept
12:21
off versus 12 pounds in the usual care
12:23
group.
12:24
So we not only took away this, you
12:26
know, drug, the GLP ones, which I'm sure
12:28
we'll cover later, but
12:30
they also
12:31
were able to lose more weight, and it
12:33
was healthy weight with an increase in muscle
12:35
mass,
12:37
which, you know, is is a really important
12:39
aspect of,
12:41
having a healthy lifestyle is making sure that
12:43
that weight loss is is a healthy weight
12:45
loss. So the study really represented a turning
12:48
point in diabetes and metabolic diseases because not
12:51
only did we show what we could do
12:53
for diabetes, but we showed what we were
12:54
doing across a host of metabolic conditions,
12:57
including metabolic associated fatty liver disease.
13:00
And it offers the strongest evidence to date
13:02
that AI digital twin technology
13:05
can really actually impact and fix disrupted dysfunctional
13:09
metabolism
13:10
without adding and actually with eliminating
13:13
expensive costly medications that have many side effects.
13:17
And just to come back to your earlier
13:19
point, I feel like this data is also
13:20
so important because, again, then
13:22
providers can go beyond what you've mentioned earlier,
13:25
which is this generalized
13:27
experience of, like, eat healthier, sleep better, which
13:30
doesn't really help.
13:31
This data backs some of those more precise
13:34
approaches up, which is so, so crucial.
13:37
You touched on GLP ones, and I do
13:39
wanna spend just a little bit of time
13:41
on that. We could spend a whole episode
13:43
talking about it, really, in this context specifically.
13:46
But they, again, they continue to create both
13:49
opportunity, but also strain for health plans and
13:51
employers. Right? There's there's cost, long term adherence,
13:54
some of those things. Right?
13:56
When you're thinking about GLP ones in the
13:59
context of your work, certainly, but also overall,
14:01
what shifts do you think are needed, and
14:04
how are you positioning Twin Health, right, within
14:07
this
14:08
evolving landscape? What what's most exciting for you
14:11
there?
14:12
Yes. So I really appreciate it. Lucas, look,
14:14
I, you know, I wanna be,
14:16
clear. I think GLP ones are an incredible
14:18
advancement in health care and in science, and
14:21
medications
14:22
play a really, really critical role in managing
14:24
metabolic disease.
14:25
And when they're clinically appropriately used, they can
14:28
absolutely help stabilize,
14:30
disease progression,
14:32
improve disease progression, manage the disease, and help
14:35
with early progress.
14:37
But I have a statement I love to
14:38
say, which is GLP ones are not for
14:41
everyone, and they're not forever.
14:43
And I think what's really important is understanding
14:46
that these medications also have a lot of
14:48
side effects.
14:50
You know, from the hair loss to the
14:52
nausea and the GI,
14:54
issues that people experience
14:56
to even just, you know, a lot of
14:58
my patients tell me this apathy toward food
15:01
and toward sharing and, you know, breaking bread,
15:04
so to speak, with their families. And,
15:07
you can also see pancreatitis, hospitalization. So there's
15:10
there's a lot to consider, and we need
15:12
to have a solution for people who either
15:14
don't want to take and now there are
15:16
orals, but there are many injectables of these
15:18
medications,
15:20
who don't wanna take these medications for life
15:22
or really don't even wanna start them, want
15:24
to try to figure out a way to
15:26
be medication free.
15:27
And so I think it's really important to
15:29
have solutions that do that. The other problem
15:32
is that
15:33
if you look at the history of diabetes
15:35
medications,
15:36
they're not actually fixing
15:38
the disrupted metabolism. They're not actually fixing insulin
15:41
resistance,
15:42
and some of these things that are at
15:43
the precursor
15:44
to the conditions that are surfacing,
15:47
with, you know, the clinical metrics that we
15:49
see or clinical data.
15:51
So Twin uses medications as one tool, but
15:53
what we are really doing is helping our
15:56
clinicians
15:57
and
15:58
patients or members
16:00
understand their own personal biology.
16:03
Watching their app every day and saying, wow.
16:05
My digital twin just said that when I
16:08
eat rice,
16:09
I have this response. But when I eat
16:11
bread, I have a very different response. Mhmm.
16:14
And, you know, this is something we see
16:16
in couples and partners who use our program.
16:19
That one person to the next, you could
16:20
be living in the same house, but you
16:22
have a very different
16:24
personalized response to the food you consume,
16:27
the order in which you eat the foods
16:28
on your plate,
16:30
and whether you slept well last night, your
16:32
stress level,
16:33
or whether you went for a walk this
16:35
morning with the dog or your son or
16:37
daughter took the dog out. So people have
16:39
a completely different response to the same exact
16:42
food just based on what's happening in their
16:45
lives.
16:46
And this is, I think, something that medications
16:48
don't really crack the nut on. And if
16:50
you want a world where people can have
16:53
no side effects, can really fix the underlying
16:56
problem,
16:57
they have to understand their personal body. They
16:59
have to understand what's precise and what's meaningful
17:02
to their health,
17:04
to their blood sugar, their weight gain, their
17:07
stress levels, their sleep.
17:09
So we really use that AI digital twin
17:11
to to monitor that and give precise recommendations
17:15
real time using that sensor data and member
17:18
preferences to the member every single day. And
17:21
that's what managing your diseases is really about.
17:23
It's about daily iteration. Right?
17:27
If you go on vacation for a week
17:29
and then you come back, you wanna get
17:31
back on track and you can do that
17:33
if daily you understand or have insights into
17:36
your own body, and that's what we're really
17:37
shifting the paradigm on in metabolic care.
17:41
We're really allowing people this chance to say,
17:44
well, I took a GLP one, but I'm
17:46
I'm done with it, or I've plateaued, and
17:48
I'm not
17:49
responding anymore, but I'm so terrified to get
17:51
off because I hear about this rebound waking,
17:55
and I don't wanna be back where I
17:57
am. So is there a solution
18:00
that can help me responsibly
18:03
have this medication, improve my health holistically,
18:06
and give me an opportunity to get off
18:08
of it? And is there a solution for
18:10
people who don't even want to start? And
18:12
that is, for me, really exciting. I mean,
18:15
my personal,
18:16
hope is that people have choice. And I
18:19
think,
18:20
health care today
18:22
isn't set up to give people choice in
18:24
their health, and Twin really gives that to
18:27
people.
18:28
I want the med, but only for a
18:30
short time, and then I want to get
18:32
off of it. And that's that's totally reasonable.
18:34
And I I think it's important to have
18:36
more and more solutions that can support that.
18:39
Yeah. I'm so glad that you mentioned that
18:40
because, again, we touched on the engagement earlier
18:42
in the conversation. But, again, that aspect of
18:44
giving somebody
18:45
agency
18:47
is truly what then enables
18:49
them thinking, okay. I can make a change
18:52
here. I can do this together, which is
18:54
so crucial. I'm so glad that you've mentioned
18:56
that. It's it's very, very critical. Doctor Shah,
18:58
it's so fantastic to have you on. Thank
19:00
you so much for taking some time for
19:01
us. Anything else that you think
19:03
deserves a little bit more attention that we
19:04
haven't touched on in our conversation here? Just
19:06
wanna turn the floor over to you. Great.
19:09
Well, thank you. Well, I'll just leave you
19:10
with one thing. I mean, we talk a
19:12
lot about AI in health care,
19:14
and I know it's everywhere,
19:17
you know, and I I see it everywhere.
19:19
But most of the talk about AI is
19:21
how do we make health care systems and
19:23
people more efficient.
19:25
Coding,
19:26
you know, how do we get them to
19:28
document more efficiently? How do we make providers,
19:30
you know, able to do their jobs better,
19:32
prior authorizations?
19:34
Twin is doing all of that absolutely with
19:37
our care teams, but we're doing it to
19:40
reengineer metabolism.
19:41
We're using AI to really improve how people
19:44
understand their own health,
19:46
how they can address
19:48
the things they struggle with every single day
19:50
and redefining
19:52
how we could
19:53
solve chronic disease and redefining the standard of
19:56
care with AI digital twin. So I think
19:58
it's really exciting, and I I hope we
20:00
can see more progress on using AI
20:03
toward the diseases and toward the conditions,
20:06
as well in the future.
20:08
Doctor Shah, so great to have you. Thanks
20:10
so much for being here. You as well.
20:12
Thank you for having me. We also want
20:14
to thank our podcast sponsor, Twin Health. You
20:16
can tune in to more podcasts from Becker's
20:18
Healthcare visiting our podcast page at beckershospitalreview.com.