Subscribe
Share
Share
Embed
Where we share our weekly news debriefs and discussions with industry experts. These are lo-fi recordings aimed at giving our readers more opportunities to engage with our analysis and a view into some of the conversations that shape it.
Martin: We're really excited
today to welcome JD Friedland.
He's the executive director for
ventures at Cleveland Clinic.
You may have heard of it.
Uh, here he is.
JD, welcome to the show.
How are you doing today?
JD: Good morning.
I'm doing great, and I'm really thrilled
to be, uh, part of this discussion.
We could spend the next 20 minutes talking
about what you guys were just talking
about, 'cause Cleveland Clinic has a
very- ⦠specific perspective on healthy
longevity, and are the Brian Johnsons
of the world and social media going to
sort of win the day with respect to how
people are going to biohack themselves,
or is something sort of grounded
in, you know, science and research
and medicine going to win the day?
We would prefer it to be one
way, but I worry that it might
end up going the other way.
So that's, that's maybe
for a different day.
Martin: Yeah.
W- w- we'll stick a, uh, stick a,
a point in that and maybe come back
to it at the end if we have time.
JD: Sure.
Martin: You have an
interesting perspective.
I'm sure you're seeing a lot of pitch
decks right now as the sort of head
of the in-house venture capital arm
for Cleveland Clinic, and I'm sure
it's, like, all AI all the time.
And so I wanna start- Mm-hmm ⦠outside
of AI and ask you, what is interesting
to Cleveland Clinic and to you
at the moment that is, is a- at
least not a pure play AI solution?
JD: Yeah, I mean, I think we have
an interesting perspective because
I think we start from the position,
like a lot of our peers, that as a
large health system, we always have
challenges, problems, and priorities
that are re- that remain unaddressed.
We are, um, we're privileged, frankly,
to be part of a system where we're
generating a lot of proprietary
knowledge, inventions, discoveries,
know-how, data, you name it.
And so we have an opport- we have
opportunities to start our own companies,
but we also have opportunities to
contribute non-cash value as a partner
to emerging companies across healthcare.
And so we start, if we're doing it right,
we're starting really with the problem.
You know, what are the biggest burning
needs that we're seeing across our
institutes, across the enterprise?
And I think a good example of that
two years ago was, hey, this ambient
listening thing feels like it's something
that might be real, and we have a big
problem with clinician burnout and
pajama time and patients getting annoyed
because doctors never look them in the
eye anymore 'cause they're spending
all their time looking at a screen.
Maybe we should look at
solutions to that problem.
And I think we really try to
start from that perspective of
what's the pr- biggest problem?
What does the ideal solution look like?
And who is out there that matches most
closely to that ideal solution that is
willing to work with us, not just as a
customer and not just as they want our
money because they want us to invest,
but as a partner where we can take an
active role in advancing that solution.
And again, that's exactly what we did
with the ambients, and that's really
become a blueprint for us on how
we can work with, um, emerging and
established players in the digital
health spectrum to, um, be additive
as a, as a customer, as a partner,
and hopefully as an investor as well.
Kevin: JD, it's good to see you.
One of the, um, one of the things
we're listening for when we listen to
largely publicly traded companies in
the space as they're talking about AI
agenda initiatives they have underway,
is for the impact of AI on either
revenue or cost for an organization
to, to think about financial impact.
I know that is one lens on, on
impact for organizations, and
there's a whole host of other lens.
Mm-hmm.
I'd be curious to, to hear how are
you thinking about measuring impact?
Do you have examples of where you feel
like AI is, is actually driving tangible
impact- Yeah ⦠organizationally
that you're excited about internally?
JD: Yeah, absolutely.
So I can give you a few examples,
and ambients is one of them
because there's a, there's
You, you guys have probably heard
this phrase many, many times before.
You have hard ROI and soft O- ROI.
You have direct ROI, and then there's
other things that we're supposed to
be delivering as a not-for-profit
mission-driven organization.
So how are we improving outcomes
which have both a direct financial
and non-financial impact on
Cleveland Clinic, on our patients?
How are we creating an environment
that's the best place to work in
healthcare, which is one of our sort
of, you know, core, uh, principles of
ourselves as an, as an institution.
And, um, how are we m- how are
we optimizing the experience
for our patients as well?
And one of the things that's always
in the back of my mind, and I've been
talking with our peers about this as
well, is, um, what does the hospital
system of the future look like?
What is ⦠Is it gonna look
like what we have today in 10
years, in 20 years, in 30 years?
How are people receiving their care?
Um, this is already, like, morphing.
And then how can Cleveland Clinic and
our peers, who deliver great care when
the patient's in the hospital, stay
connected with those patients when
they are discharged, or hopefully they
don't have to come into the hospital
because they're able to receive a level
of care based on their level of acuity
regardless of where they are, whether
they're in their house, whether they're
traveling, whether they're, you know,
come to a clinic, or they're, you know,
in some other ambulatory facility.
Um, and we want to reserve our acute
care for those cases and those patients
that need it, but make sure to conserve
those resources in such a way that allows
us to optimize care for our patients.
Martin: I am a hypothesis that Kevin and
I sort of kick around sometimes is that,
you know, there, there's sort of lots
of concerns about massive job loss and,
and sort of replacement in healthcare.
Yeah.
A sort of hypothesis that
Kevin and I kick around is that
we're, we're not going to seeâ¦
We're gonna see similar or actually
higher employment for, for health
systems, but stuff is gonna move from
largely off-stage roles, so billing
and coding and stuff like that,
to more on-stage roles, care- Yeah
care coordination, patient
experience type stuff.
I'm curious what your view on this
is- Yeah ⦠um, seeing a lot of
the, the innovation, uh, firsthand.
JD: So, A, that's already happening.
So there's a few things that I'm
trying to sort of process in my
brain, uh, you know, really in 2026.
One is, and I just came from a conference
and, you know, shameless plug, K Ventures
sponsored something called The Line Forum
last week up in Cape Cod and had, you
know, uh, Anthropic and, and, uh, OpenAI
and Walmart and many of our peer ins-
uh, uh, groups and the like, and one
of the things that emerged from that is
something that we've read in the papers.
However bad you think that, um,
general population sentiment is
on AI, it's actually a lot worse.
And so when we think about, like, ways to
help solve some of our biggest problems
in healthcare by using AI, and I'm really
happy that, by the way, we're now six
minutes into our discussion, and this
is the first time I'm using those two
letters together, uh, that was purposeful.
Um, but the consumer patient, uh,
perception of AI, not just in healthcare,
but in general, is decidedly negative.
And I saw some data at that, uh, at that
summit, which really, really shocked me.
I mean, I knew it was bad.
I had no idea how negative the general
sentiment is about AI, and AI's coming
for our jobs, and AI is going to produce
profits for tech companies, uh, at the
expense of the average person, and where
are our elected officials in trying to,
you know, put a, put a framework around
this, uh, rather than letting, you
know, the big tech companies just make
even more money and fire more people.
And that's kind of where the
general consumer mindset is.
So with, with that as a backdrop, I
do think that healthcare is uniquely
situated because it's one of a handful of
industries where there is a significant
shortage in labor, and patients feel
that because there's waiting times to see
a doctor, and I get seven minutes with
my doctor, even with Ambula listening.
Um, and so how, how can we
potentially use technology?
I won't say the two dreaded letters,
but how can we use technology in general
to, um, empower our caregivers to
deliver better care more efficiently
and create time and space for patients
to receive the care that they need?
And sometimes that's in the doctor's
office, and sometimes there may be an
ability to deliver technology to wherever
the patient is to allow them to get the
care that they need and then upscale those
patients to, uh, a caregiver when needed.
And so that's the way we're work-
really kind of framing that.
Um- I can talk about a few.
You know, I talked about
ambients a little bit.
We have partnerships with, uh,
uh, we have a partnership with a
co- with Bayesian, which I'm sure
you guys are very familiar with.
Um, we're, we're really using them
for timely detection of sepsis in
the hospital, which is a huge, huge
issue for us and everybody else.
Um, we established a partnership
last year with a company called
Dyania, which we've also invested in.
Um, and they basically are using an
AI engine to, uh, mine our patient
EMR data to identify patients first.
Uh, the first use case is to identify
patients that match the inclusion
criteria of one of more than 200
clinical studies that Cleveland
Clinic is ru- is, uh, running.
And so it accelerates our ability to
recruit patients into a study, but it
also enables one of the millions of
patients who have been to Cleveland
Clinic in the past to identify there
may be an opportunity for them to enroll
in a study for, let's say, difficult to
treat, or they may be, they may be out
of treatment options at this point, but
here's a study that they can enroll in it.
So we're enrolling patients
a lot faster in our studies.
We're now looking at ways to incorporate
that into our clinical practice as
well, and potentially identify patients
that are flying under the radar but
may have an underlying condition
that would suggest that they come
in and talk to their cardiologist or
neurologist or other specialists as well.
So I can go on and on, and we have
a partnership with Palantir with
respect to creating efficiencies.
But we really are aggressively looking
at ways to address both, I call it, you
know, back of the house and front of the
house deployment of technology to enable
us to further our mission, uh, without
building 10 new hospitals or hiring,
you know, 10,000 or 20,000 employees.
Kevin: JD, one of the topics I'm keeping
my eye on, curious about, is this
conversation around general frontier
models versus healthcare-specific models.
We saw a nature paper come out
talking about how two of the
healthcare-specific models were
outperformed by general models.
We've seen a peer institution of yours
partner on a healthcare-specific model.
Yes.
I, I'd be curious, what's, what's
Cleveland Clinic's take on, I
assume you sit on a wealth of data.
I assume that's perceived as a core asset.
How do you, how do you think about
that conversation of, of general
healthcare-specific advantages of each?
JD: So I don't know if how many of my
peer, uh, members are on the, uh, on
this, uh, event right now, but we are
in direct dialogue with investors,
with our peer institutions, as well
as with companies about best ways
to contribute, you know, our-- those
assets that you talked about and more.
So it's not just data, it's
our knowledge and expertise.
I think there's probably, uh,
uh, this might be controversial.
I believe there is one time value
to the contribution of data, and
the data that we have is never
more valuable than it is today.
Um, because that data, I believe
once, once one institution contributes
data to an LLM or a foundation
model, um, then the next set of
data becomes incrementally valuable.
But if we're the 30th institution
to contribute our data-
It's a lot less valuable than
if we're the first or second.
So we're talking with a lot of our
peer institutions about how can we
form a tight consortium of a handful
of leading institutions to contribute
really robust and deep data sets that
are complementary to one another.
Um, so that's kind of option one.
We have not announced a formal
data partnership with anybody.
We have selectively licensed some of
our data on a non-exclusive bas-basis to
inform, uh, uh, uh, foundation models,
but we're also exploring broader areas
where we can really contribute our data
in such a way that, um, is incredibly
valuable and also is a fair exchange
of value, not just for our data, but
we also want to think about once those
models are, are, um, educated, how can
they then be better informed in concert
with our expertise and our workflows?
I think that is whereâ¦
The data is a one-time value, and
there's probably incremental value of
the additional data that we generate.
But to me, the ongoing expertise and
insight that we can contribute as a center
of excellence is continuing value, and
that's more of an anu- an annuity as
opposed to, let's say, a large one-time
value that our data sets can contribute.
So that's the way I think about it today.
Um, but it also speaks to our core
function as a, um, mission-driven
organization, and this has really been
a sea change in the last three years.
Three years ago, we couldn't
license data to anyone that wasn't
a wholly owned or majority-owned
subsidiary of Cleveland Clinic.
Um, and so that was kind of
frustrating when we had companies
that we were a minority investor in
that we couldn't license data to.
We're in a very different place today.
I think we recognize the value of what
we can contribute and also the shelf
life of the value that we can contribute
Martin: We had a interesting conversation
with Zeke Emanuel a couple weeks back
about the role of clinical AI, and so
it feels like a lot of the conversation
today is focused on administrative stuff.
Makes sense.
I'm curious where you're seeing
opportunities for more, like,
AI practice of medicine or AI-
Mm-hmm ⦠clinical, like stuff
that, that falls into the clinical
bucket rather than the admin bucket.
JD: Yeah.
So, um, so I talked about the ambients.
That, to me, kind of, um, straddles
the line between administrative and
clinical, but there's definitely
a large clinical component.
Uh, again, another conference I was at
a couple weeks ago, the Peterson Health
Technology Institute summit, uh, back
in the beginning of June, they talked
a lot about this idea that, hey, I
thought AI was gonna, like, produce
efficiencies and bend the cost curve.
And if you look at the numbers, like,
expenses are up, like, 17% from,
you know, when ChatGPT kind of hit
the ground running in, in late 2023.
And I can tell you part of that reason,
I was really glad they didn't mention it
at the, at the conference, but I'llâ¦
I mean, I'll state the obvious now.
Our re- our, you know, our, um, our,
our, um, reimbursements that we're
submitting for are more robust because
we're capturing all the value that
we're delivering through ambient
listening, through other techniques.
And so, um, I think ultimately we
ought to be receiving fair value for
services rendered, and shame on us
for not being able to present a more
robust reimbursement request based on,
uh, the actual care being delivered.
But those days are gone, and now that
we have ambient listening, uh, we also
deployed, have deployed and invested in
a company called Theater that's using
ambient video to capture, um, data
coming off, visual data coming off of a
laparoscope or an endoscope, and then the
AI is able to generate a surgical note.
So now a surgeon can do the same
thing that our clinicians are doing
with ambient listening to review
and quickly confirm everything that
happened in a surgery, and then submit
that, uh, as part of a reimbursement.
And they're also losing their pajama time
too, which they're not crying over at all.
So we're looking at ways
to really deliverâ¦
Uh, again, I'm gonna stay away from
the, the two word, two-letter word,
uh, but deploying technologies on the
front of the house side to really help
create a better experience with better
outcomes, um, and reduce the burden and
the burnout factor on our clinicians.
And that's really kind of critical
to what we do and sustaining
ourselves as an organization.
Kevin: OpenEvidence has taken a,
a interesting growth strategy from
a product-led growth perspective,
getting it into the hands of clinicians
to start using its tooling, um,
organically without going through the
traditional healthcare apparatus, long
sales cycles of systems and whatnot.
And they're starting to go there,
but that's not kind of where
they started from originally.
It's, it's started to seemingly bubble
up this conversation about the shadow
use of AI inside of organizations,
and how organizations are thinking
about when, when clinicians are using
non-approved tooling, what does that,
that mean for the broader system?
I, I'd be interested to hear,
are you all thinking about that?
Is that coming up in conversations
with peer institutions?
Yeah.
How are you addressing that,
if at all, organizationally?
Yeah,
JD: it's a big issue.
I, I can neither confirm nor
deny that people may be using
OpenEvidence at Cleveland Clinic on
an unauthorized basis, but there's
certainly no authorized basis today.
Um, I did have some concerns about
the business model of OpenEvidence and
the level of, I'll just call it trust,
that clinicians and researchers can
have in OpenEvidence, given that it's
basically an advertising-driven model.
And, you know, if I was a clinician, I
might have concerns about, "Hey, look,
pharma's paying for this, so why should I
trust the information that I'm getting, or
at least the prioritization of how I'mâ¦
how the information is presented to me?"
I haven't seen to date any hard evidence
that that's in fact the case, but
that's in the back of my mind with
respect to the long-term viability
of a solution like OpenEvidence.
And, uh, again, I know we're only a
little bit into it, but I would certainly
wanna- Uh, continue to monitor that.
I think some sort of a solution, um,
and we are very focused on how do
we deliver the best information to a
caregiver at, when they need it so that
they can make an informed decision.
It really gets back to, I think, Martin,
a, a comment you made earlier, and
we use a slightly different phrase.
We call it clinical intelligence, and
I stole that from one of my colleagues
at Colorado, Colorado University.
Um, but um, that's really the,
I think the way we think about.
It's kind of a co-pilot for clinicians
and researchers to be able to
present the latest information,
evidence, and information.
I think I used information there twice.
Information, evidence, and studies,
uh, at the point of care or at
the point of research where it's
going to have the greatest impact.
I think that's gonna be critical.
I actually posed a question last week.
Uh, you know, today there are certainly
issues around, um, liability and how,
you know, like, uh, there's a lot of
companies out there providing like
virtual behavioral care and chatbots
and even agents that can interact
with patients for behavioral care.
One of the limiting factors there,
as far as I'm concerned, is Cleveland
Clinic won't just, won't, um, deploy
something like that today get, in a
current iteration because what you're
effectively doing is subcontracting to a
digital health solution, and yet Cleveland
Clinic is retaining all of the liability
associated with an adverse outcome.
So between the companies and the
health systems, we need to figure
out a fair distribution of value
and liability associated with that.
I think that's probably the pathway
forward there, and that might be
the pathway forward for any forward
deployed, clinically oriented
digital health solution that's
gonna involve either referral to
or referral from a health system.
Um, so that's, that's definitely
one, uh, major issue that we're
trying to grapple with right now.
Um, I, I probably went a little bit
sideways on your question, but I think
those are all relevant issues that
we're, we're struggling with right now.
Martin: That was a very helpful overview.
I think that liability and value question
is a, is a excellent one to end on
'cause it is, I think, going to define
a lot of the conversations that you
have and that we'll all be having, um,
over the course of the next few years.
JD, thank you so much for your time today.
We're already excited to have you
back to hear more about some of
these investments you're making and
what you're seeing in the market.
Thanks for your time.
JD: Thank you as well.
Have a great day.
Martin: Bye
Kevin: Bye.
Good seeing you, JD.
Appreciate it.