Latest / Becker’s Healthcare Digital Health + Health IT / Brenna T. Loufek, Director of AI Regulatory and Quality at Mayo Clinic

Transcript

Brenna T. Loufek, Director of AI Regulatory and Quality at Mayo Clinic · Becker’s Healthcare Digital Health + Health IT

0:02
At athenahealth,
0:03
we know your ambulatory practice wants healthier, a
0:06
healthier business, healthier care teams, and healthier patients.
0:10
But the complexities of modern health care tech
0:12
make it hard for you and your care
0:14
teams to focus on what matters
0:16
most? That's where athenahealth can help. Our AI
0:19
native all in one solutions reduce administrative burdens,
0:22
streamline billing and payments, and deliver critical insights
0:25
when clinicians need it most. That means fewer
0:28
clicks, more time for patients, and stronger bottom
0:31
lines.
0:32
Practicing medicine is complex, but running a practice
0:35
can be that much simpler with athenahealth.
0:38
See how simpler is healthier@athenahealth.com.
0:43
Hello, everyone. This is Scott King with the
0:45
Becker's Healthcare podcast.
0:47
Thrilled today to be joined by Brenna Lofek,
0:50
director AI regulatory and quality center for digital
0:53
health at Mayo Clinic. Brenna, how are you
0:55
doing today? I'm doing great. Happy to be
0:58
here. Thanks so much for joining us. I
1:00
I we have a lot of big, you
1:01
know, tech questions and and dealing with health
1:03
care to get health care to get to.
1:05
But, if you can please just share a
1:06
little bit about your background and your work,
1:08
that would be great.
1:10
Certainly.
1:11
I lead AI regulatory and quality
1:14
strategy at Mayo Clinic in the Center for
1:17
Digital Health,
1:18
where our mission is to safely translate innovation
1:22
into clinical practice.
1:24
My work centers on building the infrastructure that
1:27
allows AI technologies to to be deployed responsibly.
1:31
We develop governance frameworks, ensuring regulatory compliance, and
1:35
establishing evaluation process for digital health tools
1:39
is across the entire enterprise.
1:42
My background spans both the medical device industry
1:45
and health care delivery,
1:47
which gives me this really cool dual lens
1:50
for balancing innovation
1:52
with patient safety and ethical responsibility.
1:55
From an academic perspective, I received my master's
1:58
from Johns Hopkins in patient safety and health
2:01
care quality,
2:03
and I'm currently a doctoral student at that
2:05
same institution
2:07
focusing on the intersections
2:08
of health policy
2:10
and AI.
2:11
My research is related to
2:14
the translational
2:15
practices across health care organizations
2:17
in applying laws, regulations, and standards to AI
2:21
enabled digital health technologies.
2:24
Well, that's awesome. Thanks so much for sharing
2:26
your your great background info. I would we're
2:28
gonna get a lot for your
2:30
role. I wanna ask you about specifically everything
2:32
that's going on digitally and with tech and
2:35
in health care.
2:36
First, I wanna you know, our our research
2:39
has told us that nearly half of medical
2:41
practices reporting
2:43
using AI in some capacity
2:44
last year and, obviously, remains a key topic
2:48
for health IT leaders.
2:49
From your perspective, what are the use cases
2:52
that are making a difference right now, and
2:54
how are you leveraging them in your organization?
2:57
Yeah.
2:58
So in my opinion, use of AI can
3:00
really be broken down into two key areas
3:03
within health care organizations.
3:06
The first one
3:07
are technologies that are doing administrative support related
3:11
tasks, And then the other area
3:13
would be tools that are doing clinical decision
3:16
support.
3:17
Administrative support tools are gonna be technologies that
3:20
are helping with workflow tasks within a health
3:23
care organization,
3:24
but they're not directly doing any sort of
3:27
diagnosis, treatment, cure,
3:29
or mitigation
3:30
of any diseases or conditions within patients.
3:34
Whereas the other category of tools, clinical decision
3:36
support software,
3:37
those are doing those activities.
3:40
And there are many use cases that are
3:41
making a difference
3:42
for
3:43
the practice,
3:45
providers,
3:46
and patients
3:47
in both of these categories.
3:49
In the admin support space, we see technologies
3:53
that are aiding in scheduling and communication,
3:57
which automate
3:58
previously
3:59
manual and tedious
4:01
tasks.
4:02
These are immensely
4:03
impactful because
4:05
they help not only with time savings,
4:08
but also with burdensome
4:10
tasks that
4:12
are, health care professionals
4:14
find,
4:15
annoying or not really worthwhile
4:17
or beneficial
4:18
in the actual care delivery
4:21
actions, which are the core of all health
4:23
care institutions'
4:25
mission, visions, and values.
4:27
In the clinical decision support space, we see
4:30
technologies
4:31
that
4:32
are either
4:33
directly or indirectly
4:35
assisting health care professionals
4:37
with diagnosis
4:38
and treatment of their patients. And this these
4:41
really span everything from
4:44
tools that are
4:45
leveraging information
4:46
found within the electronic
4:48
health record
4:49
to do
4:50
detection
4:52
of patients that are at risk of different
4:54
diseases or conditions.
4:56
Maybe they would benefit from some additional screening,
5:00
or
5:01
interacting with information in that electronic health record
5:03
to give recommendations
5:05
around
5:06
treatments or diagnoses.
5:09
I believe that
5:11
around sixty percent of health care professionals are
5:14
using a tool called open evidence,
5:16
which is assisting with different differential diagnoses happening,
5:20
and that is just highly beneficial for health
5:22
care professionals.
5:24
We also see more advanced technologies
5:27
doing,
5:28
really brilliant things
5:30
where they're analyzing
5:32
medical images,
5:34
signals, or patterns
5:36
to do,
5:37
early detection and diagnosis
5:40
or to automate different workflow steps,
5:43
that take a lot of time by our
5:45
health care professionals or that are not always
5:48
done
5:49
consistently
5:50
and these are tools that are interacting with
5:53
images signals or patterns
5:55
to do things like
5:58
segment medical images
6:01
and do calculations
6:03
of distance or volumetric measurements.
6:06
There are also technologies that are interacting with
6:09
patterns or signals like in cardiology
6:12
where at Mayo Clinic, we have tools that
6:15
analyze
6:16
ECGs
6:17
to flag patients
6:18
that may be at risk of different diseases
6:21
or conditions.
6:22
And so these technologies, the clinical decision support
6:25
tools,
6:26
really help move the needle on patient care
6:29
delivery, and we're really excited about these tools
6:32
at Mayo Clinic.
6:33
Yeah. A lot of those uses you just
6:35
mentioned were are incredibly interested. I I feel
6:37
like we're hearing about
6:39
new tools
6:40
and new uses every day now. Just with
6:44
your with your role and, I guess, what
6:46
you've seen and maybe even what you've heard.
6:48
Do you do you remember ever health care
6:51
having
6:52
so much cutting edge technology involved and and
6:54
how important
6:56
tech is in in this moment in time
6:57
to health care? Like, you ever heard that
6:59
before in the industry?
7:01
No. I think that the paradigm has really
7:03
changed in the last five years.
7:07
Whereas we saw
7:08
a lot of
7:10
the med tech innovation
7:12
coming out of startup companies
7:15
or established organizations
7:17
like Medtronic or Johnson and Johnson where they
7:20
were the ones really developing
7:22
the technology
7:23
often working in collaboration
7:26
with health care institutions and providers,
7:28
but they were the legal manufacturers of these
7:31
tools. They were the ones identifying the opportunity,
7:35
and being the ones legally responsible
7:37
for developing,
7:38
implementing, commercializing
7:40
these different products. But as I said, the
7:43
paradigm has really shifted now where AI enabled
7:46
technologies
7:48
are in the hands
7:49
of healthcare
7:51
professionals themselves.
7:52
They are able to identify
7:55
local problems
7:56
that could be solved or local opportunities
8:00
that could be addressed
8:01
with AI enabled digital health tools,
8:04
And they, in collaboration often with with data
8:07
scientists,
8:08
can develop these tools
8:10
and implement them into their practice.
8:12
So we have this space of booming innovation,
8:16
but also highly localized
8:19
development of technologies
8:21
that are solving problems that the health care
8:23
professionals
8:23
themselves are seeing. And this is just going
8:26
to continue
8:27
to grow with generative
8:29
AI and the accessibility
8:32
of using established large language models
8:36
to apply it to different activities
8:38
within your organization.
8:40
So in many ways, this is a massive
8:43
boom, incredible expansion,
8:45
but it's also a real shift in the
8:48
types of individuals
8:50
that are
8:51
identifying opportunities,
8:53
building tools,
8:54
testing them, and deploying them where historically we
8:58
saw a real shift,
8:59
a real separation
9:01
of those that were doing those activities and
9:03
then those that were actually
9:05
using the technology.
9:06
So this is a really exciting time to
9:08
be in health care and, helping innovators,
9:12
healthcare professionals
9:13
take tools from discovery or ideation to actual
9:16
actual implementation into clinical practice. But it also
9:20
opens up the door for a number of
9:22
ethical considerations
9:24
that we haven't thought of before, and it
9:26
also puts health care institutions
9:28
in the position of having to apply different
9:31
governance activities that they historically have not been
9:34
responsible for.
9:36
Yeah. Yeah. I'm sure you just touched on
9:37
it a little bit. But, yeah, obviously, new
9:39
challenges come with these tools and new tech
9:42
being introduced. If you if you had to,
9:43
like, focus on your top piece of advice
9:45
for leaders as they introduce this this new
9:48
technology or or, you you know, kind of
9:50
work it out, what would that be?
9:52
Yeah. So
9:54
your governance within your health care organization
9:57
should be seen as a mechanism
10:00
that accelerates
10:01
and prioritizes
10:02
innovation.
10:03
It should not be perceived as a roadblock.
10:07
At Mayo Clinic, we use a risk based
10:10
assessment framework that allows us to evaluate these,
10:14
internally developed solutions, but also vended solutions,
10:17
quickly while maintaining
10:19
rigor
10:20
for that development and testing process that is
10:23
proportionate to the risk of the tool. I
10:26
previously mentioned that in my opinion, there's really
10:29
kind of two main categories
10:30
of AI enabled technologies.
10:33
There's those that are in the admin support
10:35
space, and then there are those that are
10:38
directly or indirectly
10:39
doing clinical decision making or aiding.
10:42
We don't wanna apply the same level of
10:45
rigor around requirements,
10:47
within the software development life cycle,
10:49
specifically with how we demonstrate quantitative or qualitative
10:53
measure of safety,
10:55
efficacy, and ethical use to tools that are
10:58
being used for admin support that are, you
11:00
know, doing basic scheduling
11:02
compared with tools that are analyzing
11:05
medical images and doing early detection of patients
11:09
with pancreatic cancer. Those are two really polar
11:12
opposite examples, but you need to make sure
11:14
that the governance infrastructure that you develop
11:17
accounts for that range of diversity of technologies
11:20
that you certainly will be implementing into clinical
11:23
practice,
11:24
but you shouldn't apply the same level of
11:26
rigor for those two.
11:28
Under our chief AI implementation
11:31
officer, Micky Tripathi here at Mayo Clinic, we
11:34
are currently developing top down policies
11:37
that standardize those evidence requirements.
11:40
So our teams know in advance
11:43
what is needed for safety and compliance.
11:46
That transparency
11:48
really shortens the path from concept
11:51
to clinical use and implementation in the practice.
11:55
Innovation must move at the speed of trust,
11:58
and this is the trust with the practice,
12:02
providers,
12:03
patients, and the public.
12:05
At Mayo Clinic, we encourage experimentation
12:09
through suggested
12:10
tiered deployment
12:12
that allows for evaluating
12:14
safety,
12:15
equity, and value before scaling occurs.
12:19
This approach allows us to be bold in
12:22
trying new technologies
12:23
while maintaining the discipline that patients expect from
12:27
a world class health care institution.
12:30
And sticking with it, I think that's that's
12:32
fantastic. And sticking with
12:34
advice, you know, the the a lot of
12:37
the the use cases you work out, you
12:38
you know what's coming in, you work with
12:40
it. But the the unknown, you know, seeing
12:42
all the advancements we have had recently, like
12:44
you said, the last five years has been
12:46
monumental. But, you know, just thinking of what
12:48
could be out there down the road, what's
12:50
your top piece of advice for leaders as
12:52
they prepare for further
12:54
advancements in technology
12:55
and kinda weighing those with the rising demands
12:57
for care?
12:58
Mhmm.
13:00
I think the future belongs to those organizations
13:04
that effectively
13:05
operationalize
13:06
governance. Establish multidisciplinary review structures early and
13:09
multidisciplinary
13:10
review structures early, and ensure decisions about AI
13:14
deployment are data driven,
13:16
transparent,
13:17
and clinically grounded.
13:19
It is essential that you embed your organization's
13:22
relative risk tolerance
13:24
into this governance
13:26
structure.
13:27
What we've done here at Mayo Clinic may
13:29
not be directly applicable to your healthcare organization
13:32
because maybe you have a different risk tolerance.
13:35
Technology can't transform care unless people trust and
13:39
understand it.
13:41
So leaders should prioritize
13:43
digital literacy,
13:44
cross functional collaboration,
13:46
and clinician engagement
13:48
just as much as the technical infrastructure.
13:51
I really think that culture is the true
13:54
enabler of responsible AI
13:57
or really in any innovative
13:59
technology.
14:01
Every AI or digital tool must serve a
14:05
clear clinical or operational purpose,
14:08
and we should measure for outcomes that support
14:11
that this tool is achieving that through post
14:14
deployment monitoring activities.
14:17
Lastly, my advice is to align
14:20
innovation with your institution's
14:22
mission,
14:23
whether that's advancing healing, equity, or access for
14:27
your patients,
14:28
and then hold technology to the same ethical
14:31
standards
14:32
as the practice of medicine itself.
14:35
Thank you so much. And the the last
14:36
thing I wanna ask you is,
14:39
obviously,
14:39
all this tech that we're working with and
14:42
and applying in health care,
14:44
you and your team are, it takes it
14:46
takes great leadership. Right? So I just wanna
14:49
ask you on a human level, how have
14:51
you evolved as a leader?
14:54
So I started at Mayo Clinic
14:56
as a manager of the same space in
14:59
regulatory
15:00
and quality
15:01
where I was coming from the med tech
15:03
industry
15:05
really with the primary goal of getting technology
15:08
into,
15:10
the commercialized space. So my experience was really
15:13
in
15:14
submitting products to FDA and understanding what that
15:18
pathway to
15:19
commercialization
15:20
would
15:21
be. My formal and formative regulatory years were
15:24
at Medtronic in their cardiac rhythm and heart
15:27
failure division.
15:28
And here at Mayo Clinic,
15:30
my job is not so much to pursue
15:33
the commercialization
15:35
of these different technologies,
15:37
but to see how we can build those
15:39
roadmaps
15:40
from research
15:41
to implementation
15:43
into clinical practice in a local way. So
15:46
as a leader, I have really grown and
15:49
expanded
15:50
my understanding
15:52
of what regulatory
15:53
and quality means
15:55
when building roadmaps or pathways to implementation,
15:59
where I've really shifted my mindset
16:01
from the angle
16:02
of getting FDA clearance or approval for these
16:05
technologies that are then going to be marketed
16:08
to building pathways more generically
16:11
for broader AI enabled digital health technologies,
16:14
to be implemented in the clinical practice, but
16:17
not necessarily
16:18
commercialized.
16:19
So I've had to root myself
16:22
in the clinician's perspective
16:24
of understanding
16:26
what it really takes to develop safe, effective,
16:29
and ethical products
16:31
while leveraging information from laws, regulations, and standards,
16:34
but really relying on their expertise to understand
16:38
how and when can we develop AI technologies
16:41
to solve these local problems.
16:43
So I have I have grown from what
16:46
I would say is a pure,
16:48
formal regulatory
16:50
expert to being someone who is more of
16:53
a generalist
16:54
in translation
16:55
science. And I think that this is where
16:57
we are going to see,
16:59
opportunity
17:00
and growth in leaders
17:02
with effective implementation
17:04
of AI,
17:05
where we
17:07
build leaders and, expert contributors
17:10
who understand
17:11
the regulatory ecosystem
17:14
and what is written in laws and standards,
17:18
but they take a generalist approach that really
17:20
leans on the clinician's expertise
17:23
to understand what it takes to pragmatically
17:25
translate these tools into clinical practice.
17:29
Prada, thank you so much for joining us
17:31
on the podcast. I think it was a
17:32
great conversation. I learned a lot personally. I
17:34
know everyone listening did, and I hope to
17:36
work with you again soon.
17:37
Thank you so much.
17:42
At athenahealth,
17:43
we know your ambulatory practice wants healthier,
17:46
a healthier business, healthier care teams, and healthier
17:49
patients.
17:50
But the complexities of modern health care tech
17:52
make it hard for you and your care
17:54
teams to focus on what matters
17:56
most. That's where athenahealth can help.
17:58
Our AI native all in one solutions reduce
18:01
administrative burdens,
18:02
streamline billing and payments, and deliver critical insights
18:05
when clinicians need it most.
18:07
That means fewer clicks, more time for patients,
18:10
and stronger bottom lines.
18:12
Practicing medicine is complex, but running a practice
18:15
can be that much simpler with athenahealth.
18:18
See how simpler is healthier at athenahealth.com.