Latest / Becker’s Healthcare Podcast / Intelligent Health and Empowering People Through Data and AI with Dr. Nasim Afsar

Transcript

Intelligent Health and Empowering People Through Data and AI with Dr. Nasim Afsar · Becker’s Healthcare Podcast

0:00
This is Scott Becker with the Becker's Healthcare
0:02
Podcast.
0:03
I am thrilled today to be joined by
0:05
a brilliant physician leader, but we're joined today
0:08
by doctor Nassim Afsar.
0:10
And doctor Afsar has written a new book.
0:13
She's had a brilliant career,
0:14
and we'll talk about that. Be before we
0:17
get started on what you're doing today in
0:20
in in in in so on, can you
0:22
take a moment to introduce yourself and tell
0:25
the audience a little bit about your career
0:27
and and your brilliant career, quite frankly?
0:29
Oh, thank you so much, Scott. I'm thrilled
0:32
to be here again,
0:33
with you to talk about my new book,
0:36
Intelligent Health.
0:38
I am a a physician by training and
0:40
background. I trained in internal medicine and was
0:43
on faculty at, UCLA Health as a hospitalist
0:47
for over a decade taking care of acutely
0:49
ill patients,
0:51
but really got into health care administration early
0:53
on in my career. And kind of the
0:55
force behind that, Scott, was
0:58
I became a physician because I wanted to
1:00
make sure
1:01
that we were
1:02
creating and innovating to make large scale impact.
1:05
My way of doing that was by being
1:08
at bedside,
1:10
integrating all the information
1:11
that,
1:13
that an acutely ill person had and trying
1:16
to work with a talented team of physicians,
1:19
nurses,
1:21
and staff to figure out how we can
1:24
help that person,
1:25
transition towards healing.
1:27
Well, when I started to see the gaps
1:29
that we had in health care
1:31
at the beginning of my career, I became
1:33
very interested in how do we actually build
1:36
more robust systems of health and care and
1:39
got into the c suite at UCLA as
1:41
their kind of inaugural,
1:43
associate chief medical officer, which was this was
1:46
in the early days, so this was the
1:47
chief
1:48
the chief quality officer
1:50
of two decades ago,
1:52
before we kind of had a title for
1:54
that. And I went from bedside where I
1:56
was taking care of 20 people to
1:59
now thousands of people through our quality work,
2:03
at UCLA. I was then asked to take
2:05
on a broader role in population health management,
2:08
helping people
2:09
get better health and care from birth through
2:11
the transitions at the end of life, gave
2:13
me an opportunity to care for hundreds of
2:15
thousands of people,
2:17
became very interested in how do we actually
2:19
finance
2:20
and,
2:21
and move things forward in
2:24
an effective
2:25
way that's sustainable for the long term.
2:28
Got my MBA and became a chief operating
2:31
officer at UC Irvine Health,
2:33
where, again, I now had an opportunity to
2:35
impact millions of people.
2:37
And then my,
2:39
very first,
2:41
CEO, one of my, wonderful friends and mentors,
2:44
David Feinberg,
2:46
recruited me to,
2:48
Cerner as their inaugural chief health officer,
2:51
with the idea that technology is our most
2:55
powerful lever
2:57
to be able to advance health and care
2:59
across the globe. And we had,
3:01
over 300,000,000
3:02
people across the globe who counted on our
3:04
technologies for better health and care. Well, you
3:07
know the the story with that, we got
3:08
acquired by Oracle within six months of me
3:11
joining,
3:12
and then Oracle retained me as as their,
3:15
chief health officer.
3:17
And,
3:17
at Oracle, I got to
3:19
work with health care delivery systems across the
3:22
globe, but also major payers, pharma, medtech,
3:27
pharmacy retail,
3:28
food and beverage companies,
3:30
Scott, because as you know, everyone is trying
3:32
to figure out how do we kind of
3:33
leverage technology to get the health outcomes that
3:36
we need as well as,
3:39
create operationally and financially sustainable organizations.
3:43
Love that position, transitioned out about a year
3:46
ago to write Intelligent
3:48
Health, which is kind of the evolution of
3:51
where I've come, where my focus is really
3:53
on how do we leverage these technologies
3:56
to empower
3:57
the consumer of health and care to thrive.
4:00
That brings us today.
4:01
No. And thank you. Thank you. So you've
4:03
worked at these major institutions,
4:06
from UCLA Health
4:08
to Oracle, Cerner,
4:10
and and now your focus in in in
4:12
is now the new book, Intelligent Health,
4:15
really centers on the individual.
4:17
Yeah. Talk about what prompted that
4:20
that sort of pivot
4:22
to the individual versus the macro macro systems.
4:25
Talk about that in the book Intelligent Health.
4:28
Yeah. Absolutely. So, you know, it's interesting, Scott.
4:31
As I reflect,
4:32
about the
4:34
last two decades of my career,
4:37
I have realized that
4:39
this focus on the person,
4:42
was always something that was
4:45
near and dear to my heart and part
4:47
of my core mission. So back again, going,
4:50
you know, two decades back,
4:52
when I was put in charge of our
4:55
health system readmission program, I actually created a
4:57
program
4:58
that was called person centered readmissions.
5:01
And the reason that I did that was
5:03
that we had done extensive chart reviews
5:06
and found that despite significant
5:09
data
5:10
from the literature on what causes
5:13
readmissions, like
5:14
lack of follow-up with a primary care doctor
5:17
or,
5:18
medication
5:19
reconciliation
5:20
issues, that those weren't really the factors that
5:23
were
5:24
leading to readmissions with our patient population. And
5:27
I remember I kinda got this data from
5:28
our analyst at about 6PM one night. I
5:31
was sitting in my office,
5:33
at UCLA, and I was like, well, what
5:35
does cause readmissions? And printed out
5:38
the list
5:39
of patients who had been readmitted overnight and
5:42
walked onto
5:44
the floor at UCLA and just started talking
5:46
to to patients, introduced myself,
5:48
asked them why they came into the hospital.
5:50
And this may shock you, Scott, but I
5:52
found that, again, for our patient population, it
5:54
wasn't that they didn't have a primary care
5:56
doctor or a specialist.
5:58
It wasn't that their medications
6:00
interacted
6:01
negatively and that's what led them to come
6:03
in. It was that they felt safer in
6:05
the hospital.
6:06
So if we wanted to prevent readmissions,
6:09
we had to address
6:11
their sense of safety when they were transitioning
6:14
home. And so we rebuilt, and that's why
6:16
it was called person centered readmissions. We rebuilt
6:18
the whole program
6:20
to do an assessment before discharge
6:22
around how comfortable and safe do you feel
6:25
at home. And when people didn't feel comfortable
6:27
and safe, what were those factors, and how
6:29
could we help address that? And that was
6:32
a thread that went with me. I set
6:34
up
6:34
some of our first patient family advisory councils,
6:38
and I realized
6:39
as I went through my career, especially as
6:41
I became a chief operating officer and then
6:44
when I worked globally through Oracle,
6:47
that you have to really
6:49
understand
6:50
what does the person or consumer of health
6:52
and care need. And those needs really vary
6:55
from person to person even if they are
6:58
within the same ZIP code or geography, and
7:01
you have to ultimately be able to be
7:03
able to address that. And then that brings
7:05
us to intelligent health.
7:07
And and and talk about that because, obviously,
7:11
at the end of the day, there's these
7:12
macro concepts. There's population health. We worked a
7:14
lot on population health, but but then there's
7:16
actually the very real individual. Is that individual
7:20
moving enough? Are they going to the doctor
7:22
when they need to go to the doctor?
7:23
Are they drinking the water they need to
7:25
drink? Are they eating healthy? Are they eating
7:27
not healthy? Are they taking their medicine that
7:29
they're not? How do you get from this
7:32
macro level
7:33
to to really working the individual on the
7:36
individual
7:37
level? It's an inspiring goal. What does that
7:39
look like in practice? And and how can
7:41
then we tangibly improve
7:44
a person's health by really looking at it
7:45
from this micro level, the individual person,
7:48
versus the macrosystem level?
7:50
Yeah. That's a really great question, Scott. And
7:53
the the really exciting thing about sitting here
7:55
in 2026
7:57
versus,
7:58
you know, again, looking a decade ago or
8:00
two decades ago is that technology is now
8:03
an enabler
8:05
to help us
8:06
have that personalized
8:09
precision,
8:11
health and care path
8:13
for the for the individual.
8:15
So the idea behind intelligent
8:17
health
8:19
falls on three kind of core premises.
8:22
The first one is we have to unify
8:25
data.
8:26
We have to unify the 20%
8:28
that determines someone's health, which is what happens
8:30
within health care delivery systems,
8:32
with the 80% that determines their health, which
8:35
is, as you know, Scott, the air we
8:36
breathe, the food we eat, the stress we
8:39
carry, our genetics, our habits.
8:41
It is unbelievable
8:43
to me
8:44
that we make
8:46
such critical decisions about health and care with
8:49
only 20%
8:50
of the information.
8:51
You wouldn't get on a plane if the
8:53
pilot came on and said,
8:55
thank you for,
8:56
flying with us today. Just wanna let you
8:58
know I only have 20% of the data
8:59
to get you to the destination safely, but
9:02
I'm highly trained and very well intentioned. I
9:04
mean, you would you would get off the
9:05
the the plane. And yet when it comes
9:08
to our health care, we are focused on
9:10
the 20%,
9:12
leaving the core
9:14
elements that determine our health to the side.
9:16
Well, now we have the technologies to be
9:18
able to unify
9:19
that 20 with the 80 and for the
9:21
first time kind of see the whole person.
9:24
The second,
9:25
pillar of intelligent health is we now can
9:28
apply intelligence so that with all of the
9:31
these streams of data that we have from
9:34
our
9:35
wearables to apps to genetic testing that we
9:38
do, full body MRIs,
9:40
testing
9:42
of lots of blood panels.
9:44
Along with our electronic health record, we can
9:46
actually distinguish
9:48
signal from noise. So we can understand
9:50
and have insight into what drives health and
9:53
care today
9:54
and foresight about what's gonna drive our health
9:57
and care in the future.
9:59
And that enables us,
10:01
to to have the ability to make different
10:03
decisions today. And I I have a fun
10:06
story about that that I'll I'm happy to
10:08
share with you later.
10:09
And then the last pillar of intelligent health
10:12
is this capability, this unified data with a
10:15
layer of intelligence has to be
10:17
person owned and person driven.
10:20
Because when you own your data,
10:23
when you have control over it, you are
10:25
going to fundamentally
10:27
think and act differently, and you're gonna have
10:30
human goals. We've kind of trained people. Like,
10:32
if I ask you what are your health
10:34
goals, you may say, you know, I need
10:36
to control my cholesterol more. I need to
10:39
lose a couple of pounds. These are clinical
10:41
goals. We've trained people to think about health
10:44
in a clinical way.
10:45
The thing is that we need to think
10:47
about health in human ways. I want to
10:49
wake up in the morning and
10:52
have energy throughout the day.
10:54
I want to go from meeting to meeting
10:56
with a clear head. I want to be
10:58
able to
11:00
run with my kids and not have musculoskeletal
11:03
pain. The the these are human goals. The
11:05
intelligence on the background
11:07
can take the twenty percent of your chronic
11:09
illnesses
11:10
with the eighty percent
11:12
of your Lyft contacts and tell you on
11:14
a day by day, minute by minute basis,
11:18
how can you make better decisions to get
11:20
to those goals?
11:22
So so thank you. And I think that's
11:24
right on. I mean, in so many different
11:26
pieces of life that we have, the whole
11:28
game is to be able to show up
11:30
present and alert and feeling good and to
11:33
be able to move how you want to
11:35
move
11:36
and to be, you know, ready to go
11:38
and raring to go, to sleep well, to
11:39
feel good, all these things. Talk for a
11:42
second about artificial intelligence.
11:45
You know, there there's one school of thought.
11:47
You see it on Twitter often or x
11:49
that AI is gonna take over everything. We're
11:51
not gonna need doctors.
11:53
Most of us, I don't think, view it
11:55
that way. We we're very concerned that we
11:57
have enough specialists, enough doctors, enough people take
11:59
care of us.
12:00
Where does AI fit into all this and
12:02
and and and and how will it be
12:05
useful, and what are the limitations?
12:08
Yeah. I think that there's kind of two
12:10
major areas where AI is gonna be absolutely
12:13
instrumental. I think the first one is to
12:15
take away
12:16
the tasks from the person that that should
12:20
have never been done by a human being.
12:22
There are,
12:23
as you know, tremendous administrative tasks that highly
12:27
trained physicians, nurses,
12:29
and even staff do,
12:31
that honestly, we no longer with today's technology,
12:34
we no longer need a human to do
12:36
those tasks. We should take that away so
12:38
that we are freeing up time and cognitive
12:41
bandwidth for people to actually do the human
12:44
work, which is connecting,
12:47
communicating, building trust, actually being able to do
12:51
complex decision making that relates to the lived
12:54
context of the person in front of them.
12:57
I think the second way that AI can
12:59
be absolutely helpful is where it can help
13:02
augment
13:03
the work that's already being done. So we
13:05
know, Scott, that we have long surpassed the
13:08
cognitive ability of clinicians to integrate
13:11
all the data that's out there that's continuing
13:14
to grow and,
13:16
and develop.
13:18
We've surpassed the ability to cognitively bring all
13:21
of this stuff together
13:23
and, be able to
13:25
integrate it in the right statistical way for
13:28
the person in front of us in clinic
13:30
or in the hospital,
13:32
AI gives us the ability to do that.
13:34
It gives us the ability to pull all
13:35
of that together, the latest and greatest data.
13:38
It gives us the ability to pull real
13:39
world evidence,
13:41
which we would not otherwise have access to.
13:43
And so it enables
13:45
us to augment the human.
13:48
It's similar too with,
13:50
clinical documentation now and being able to take
13:52
some of that away from nurses and physicians
13:55
and having systems actually do that, again, augmenting
13:58
the human so that you're you're building more
14:01
bandwidth. So I think, you know, those are
14:03
two critical areas where it can be very
14:06
helpful.
14:07
I do not believe that AI is going
14:09
to,
14:11
replace health care providers. I absolutely believe that
14:15
it is going to require them
14:17
to think and function differently, and that's a
14:20
good thing. I didn't get into medicine to
14:22
practice
14:23
the way people did twenty years ago or
14:25
fifty years ago. People today should not be
14:28
wanting to practice medicine the way we did
14:30
ten years ago or twenty years ago. We
14:31
need to grow and evolve
14:33
with these new tools. But your point about,
14:36
you know, what are the what are the
14:38
downsides,
14:39
again, I think there there are,
14:42
two that are very top of mind for
14:44
me. One is around
14:48
these,
14:49
generative AI models are ultimately
14:52
trained on existing data
14:54
and are only as good as the existing
14:57
data that we have. And we know that
14:58
there are major gaps that we have
15:01
with,
15:02
with training data. We have gaps in the
15:04
training data when it comes to things like
15:06
women's health or with mental health, and these
15:10
are areas where,
15:12
these LLMs can really fall short. Again, not
15:15
because they're malicious,
15:16
but because they haven't been trained appropriately
15:20
for the the task ahead. So we have
15:22
to be very cognizant of,
15:25
is the question that I'm asking, is the
15:27
augmentation that I'm requiring,
15:30
appropriate
15:31
for the model that I'm using? And then
15:33
I think the second piece that I that
15:35
I really think about and and,
15:37
worry about
15:39
is around,
15:41
as we use these systems more and more
15:44
as we are deferring to them, are we,
15:48
continuing to build and develop our critical thinking
15:52
skills, our abilities to discern
15:54
what is right and wrong? And I do
15:56
think that that's a very human,
15:58
skill that's critical to continue to build. And
16:01
there was a, study out of MIT,
16:04
where they asked a group of adults to
16:07
write an essay. And they asked,
16:10
for the for the group that was just
16:12
told to write an essay. They then did
16:14
EEGs and kind of followed their brain patterns.
16:16
Their brain was firing.
16:18
For another group, they said, write an essay,
16:20
but you can use Google as a search
16:22
engine.
16:23
The brain for those folks were firing, but
16:26
less. And then they asked a group to
16:27
write an essay, but leverage,
16:30
a large language model, and the brain there
16:33
was minimally firing.
16:35
Then they asked people to recall
16:38
something from their essay. And, again, if you
16:40
wrote it yourself, you recalled it. If you
16:41
used ChattGPT,
16:43
recall was really poor. They ask people how
16:45
satisfied they were with their work, their level
16:48
of happiness. If you wrote the essay, if
16:50
you toiled over it, you were actually you
16:52
had more fulfillment and happiness over your work,
16:55
all the way to if you leveraged an
16:57
LLM, your level of joy, happiness, fulfillment went
17:00
down. So I I bring that up because
17:03
I worry about at a time when we
17:05
have burnout and workforce shortages,
17:08
that how we use these have an impact
17:11
in how we feel and how we perceive
17:13
our professional work. And we have to be
17:15
mindful that it's not,
17:17
further
17:18
eroding
17:19
at the workforce
17:21
that is already,
17:23
you know, quite taxed.
17:25
Thank you. And and and the point you
17:27
make, I mean, there's so many points there.
17:30
One is that
17:32
I I hear this from teachers today
17:34
that as students rely so heavily on
17:38
AI and chat GBT,
17:40
well, their work product may be good, their
17:42
ability to actually rigorously think is not being
17:45
improved nearly as much. And that is a
17:47
real, real challenge. The other thing you said
17:49
that I think is is so right on,
17:51
I see that a different context in the
17:53
oncology world
17:55
is just the level of cognitive
17:57
overload
17:58
that oncologists are facing at a time when
18:00
there are shortages and more people with cancer
18:02
and so forth.
18:03
And trying to deal with that cognitive overload
18:05
and filter through so much information
18:08
to be effective as a physician,
18:10
and even though the tools are so helpful,
18:12
there still is this incredible challenge of being
18:15
able to avoid burnout,
18:17
maintain
18:18
empathy, think clearly, and filter through thoughts in
18:21
a time of tremendous
18:23
cognitive overload.
18:24
Absolutely.
18:25
Absolutely, Scott. I just had a it's interesting
18:28
that you bring up that case. I had
18:29
a good friend of mine, reach out the
18:31
other day,
18:32
and,
18:33
she is now,
18:35
six years cancer free, from breast cancer, was
18:38
seen at one of the top three
18:40
cancer institutes in The US,
18:43
and,
18:44
it was just a regular annual visit.
18:47
She had to wait over ninety minutes in
18:50
clinic to be seen,
18:52
and there were a number of issues that
18:54
she needed to go through. She had all
18:56
of three minutes with her oncologist.
19:01
And, again, there there was a lack of
19:03
empathy and understanding for the fact that she
19:05
had moved her entire day around to actually
19:09
be present. And I think a lot of
19:11
people encounter this. And, again, we have surpassed
19:14
the cognitive ability for people to pull stuff
19:16
together.
19:17
It's causing exhaustion and burnout, which really gets
19:20
at compassion and empathy for the person in
19:22
front of us. And there are solutions
19:25
through leveraging technology, not
19:27
as a panacea, but as a powerful tool
19:29
to help us move forward. But but I
19:31
think your point is so well well taken
19:33
that it's not a panacea. We need both.
19:34
We need technology, and we need doctors. No
19:36
time to think,
19:38
to treat, to think, to actually
19:40
to avoid some of the cognitive overload of
19:41
what you're talking about. It's that concept if
19:43
you get three minutes with the physician, it's
19:45
a very important diagnosis,
19:46
and that physician spends three minutes with you
19:48
because they're trying to sort through all this
19:50
different different places that's supposed to simplify it.
19:53
But humanly, it's just not possible to work
19:55
through that in the right way
19:57
without having a little bit more time and
19:59
and and a chance to decompress and to
20:01
think clearly.
20:03
When you think about intelligent health and what
20:05
you write about in person centric care,
20:08
What does this look like over the next
20:09
five to ten years? What shifts do you
20:11
see? How do you see this if it's
20:13
working out well? And then then take one
20:14
more moment at the end of this, doctor
20:16
Afra, at the start, to tell people where
20:18
they could find the book, Intelligent Health.
20:20
Yeah. Absolutely.
20:22
So,
20:23
you know, if we are doing this
20:26
right, Scott, what five to ten years from
20:29
now will look like is individuals
20:32
empowered
20:33
and engaged in their health, not in their
20:36
illness. And, again, there will be there will
20:38
be illness, but this will be part of
20:40
our health and how we are moving and
20:42
advancing. We have the tools. We have the,
20:45
the,
20:46
empowerment to actually be able to thrive. But
20:49
what's really critical about this, and I, spend
20:52
an entire section in the book kind of
20:55
going chapter by chapter through,
20:58
the evolution
20:59
that
21:00
health care delivery systems need to take,
21:03
the payers and employers,
21:06
in in health care, the,
21:09
pharmaceutical
21:10
companies,
21:10
medtech,
21:12
retail pharmacies,
21:14
food and beverage companies,
21:16
regulators and policymakers.
21:18
I talk about how each of these organizations,
21:21
instead of functioning in silos
21:24
and at times competing with each other or
21:26
creating friction,
21:27
how they will evolve
21:29
around the consumer and ensuring the consumer's health.
21:32
I specifically address
21:34
how do they each preserve their individual mission
21:38
and their margins, but around health as opposed
21:42
to supporting illness or contributing to illness in
21:45
in some cases. So if we're doing this
21:47
right, the entire ecosystem of health and care
21:49
because we we have
21:51
a large, complex ecosystem. We can't blow up
21:54
any of these
21:55
stakeholders.
21:56
We are too interconnected.
21:58
But how can we evolve all of them
22:01
leveraging
22:02
this technology
22:03
to,
22:04
be around the only constant that they all
22:06
share, which is the individual, the consumer of
22:08
health and care, and how can they focus
22:10
on the on the health of the individual.
22:13
All of this is explained
22:14
in-depth in my book, Intelligent Health, the Movement
22:17
to Unify Data, Harness AI, and Empower People
22:20
to Thrive. You can pick up a copy
22:23
anywhere you get your books and, and ebooks
22:26
and,
22:27
would,
22:28
love to hear any thoughts and dialogue that
22:31
you have. You could follow me on on
22:33
LinkedIn.
22:34
But this was a lovely conversation, Scott. So
22:36
happy that we got a chance to catch
22:38
up. Doctor Absa, just a pleasure to visit
22:40
with you. Thank you so much for joining
22:42
us today on the Beckers Healthcare podcast. It's
22:44
a remarkable career. It's remarkable to visit with
22:46
you. I look forward to digging into intelligent
22:49
health. Thank you so much for joining us.
22:51
Thank you, Scott.