September 24, 2025 • 29 mins
AI is bringing sweeping changes to drug development, from how targets are discovered to optimizing clinical trials to maximize an asset’s chance for success. On a special edition of the BioCentury This Week podcast, IQVIA’s Greg Lever joins BioCentury’s analysts to discuss agentic AI’s short- and long-term prospects to help biotechs discover new targets, predict success in preclinical development, and enhance clinical operations. This episode of BioCentury This Week is sponsored by IQVIA Biotech.
View full story: https://www.biocentury.com/article/657086
#Biotech #Biopharma #DrugDevelopment #ClinicalTrials #TargetDiscovery #AgenticAI #GraphRAG #DeRisking
00:01 - Sponsor Message: IQVIA Biotech
01:22 - AI in Biotech
05:01 - Machine Learning
06:21 - Generative AI and Language Models
08:37 - Agentic AI
12:43 - AI in Target Discovery
23:44 - AI in Clinical Trial Design
To submit a question to BioCentury’s editors, email the BioCentury This Week team at podcasts@biocentury.com.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
[AI-generated transcript.]
Eric Pierce (00:02):
BioCentury This Week is brought to
you by IQVIA Biotech.
For biotech companies strivingto bring innovative therapies
to market and maximize patientimpact, IQVIA Biotech is
the trusted CRO of choice.
Backed by 25 years ofunparalleled experience
and deep therapeuticexpertise, our full-service
clinical developmentsolutions are purpose-built
(00:23):
to accelerate success.
IQVIA Biotech helpsearly-stage biotechs de-risk
by developing strategicclinical development plans,
guiding drug candidates alongthe most promising pathways.
Leverage data-driven modelsand dynamic tools to craft
a compelling value story andmaintain momentum through every
phase of drug development.
Jeff Cranmer (00:49):
AI is the buzzword everywhere these days and
biotech is no stranger to that.
Welcome to a specialedition of the BioCentury
This Week podcast.
Today I'm very pleased.
we have a special guest fromour sponsor, Greg Lever,
he's the Director of AISolutions Delivery at IQVIA.
(01:13):
And joining me as wellare my colleagues and
podcast, regulars, SelinaKoch and Lauren Martz.
And we're gonna cut rightto the chase, Selina, when
you think of AI in biotech,uh, what does it say to you?
Selina Koch (01:29):
right now if I try to take a step back and
like take in all these changesthat are sweeping through the
industry so rapidly, it kind ofmakes my head spin a little bit.
conversations about, howto use AI and agentic AI and
drug development, are now,you know, being happening
in, so it seems like in everysingle company, at every
(01:50):
investment firm, and it's even.
Maybe even hard to go toa meeting where the topic
doesn't come up right?
So things are changing atkind of dizzying speed, but
because of that, we're in thislike moment where there's this
wide variability across theindustry in terms of how much
or little people understandabout the technology and
how proactive they are inengaging with it, I think.
(02:13):
And so I think it'sa really nice time to
have this conversationto kind of take stock.
Of where agentic AI is,making some practical inroads
and, where it has, otherapplications in the future.
if I could say one morething, I don't think that's
always obvious, like where AIis gonna make contribution.
like I don't think manypeople would've guessed
that writing code would beamong the first breakthrough
(02:35):
applications of AI that.
First jobs under threat would beprogramming jobs, for example.
But you know, in retrospectit makes sense when you think
about the data sets thatare available in that space.
There are, well, I don'tknow, or Greg might know
this, um, tens of millionsof programmers who regularly
check code into GitHub.
So there's this really bigrepository that not only has
(02:58):
a lot of code, but has like acode progression, leading to
code working or not working.
So you have.
Also this referenceto ground truth.
but when I think aboutbiology, it's really not
that straightforward, right?
We have this like fragmentedevidence ecosystem.
We have a few large scale datasets would have that kind of
reference to ground truth.
(03:18):
that makes me question in thenear term where there's gonna
be practical application.
Like if you think ofprotein design, for example.
I can imagine that optimizationof antibodies and proteins
along certain parameterswill, will be happening.
It is happening, right?
But wholesale de novo designcould take a while longer.
I dunno.
If you all disagree,feel free to chime in.
(03:39):
But
Jeff Cranmer (03:40):
Lauren, do you have any thoughts
that you wanna throw outthere before we, uh, go?
I see Greg nodding alongand we'll, we're gonna
bring him in in just aminute, but, uh, Lauren.
Lauren Martz (03:48):
Yeah, I think we can pass it to Greg momentarily,
just to follow up on some ofthe ways that this technology
might be finding its way intoour industry and already is.
there's protein design,there's target discovery.
And that goes all the waythrough to ways that, AI can
support clinical trial design.
you know, we've been hearingabout this for years.
(04:10):
I think a lot of what we'veheard initially is that.
This can be used to help withsite selection, you know,
with, finding the locationsand all the different
dimensions that go into that.
This is a place where datacan have a big impact.
You know, where the patientsare, where the right
investigators are, um, youknow, how many trials are being
run in that, in that site.
This is something that we,we've already seen this in
action, from Greg, I'd loveto hear more about how AI
(04:34):
can support clinical trialdesign and everything in this
spectrum of drug discovery,through drug development.
Jeff Cranmer (04:41):
Alright, well let's bring in Greg.
It's Greg Lever, director ofAI Solutions Delivery at IQVIA.
And Greg, I know, we'vechatted a bit before and
I think you thought it wasimportant to kind of set some
definitions to get us oriented.
why don't I hand it over to you?
Greg Lever (04:59):
Thank you.
Yeah, thanks for having me.
It's great to be here.
Yeah, and I think to Selina'spoint, there's a lot going on
in the industry and I thinkit can be quite difficult
to get a, a proper grip on alot of the terminology that's
out there and the differentapproaches and, and also why
they matter and, and, andwhat it is that we can, we can
actually, you know, use here.
So I think to act as a, astarting point, it can be,
(05:21):
helpful to think about.
How all the different termsinterconnect is specifically
aspects like ai, machinelearning, generative ai,
ai, and if we start with ai,as a broad term, artificial
intelligence is really, youknow, used to describe all of
these sorts of initiatives.
And it was, started asan academic discipline
(05:41):
in the mid 1950s.
and there are now vastdifferent types of
approaches and algorithmswithin this space and.
One of these approachesis, machine learning.
So this is where, you know,typically you have algorithms
that can be used to train amodel so that it can learn from
data, make predictions basedon unseen, but related data.
So, as a more concreteexample, say I have data on
(06:05):
adverse events that occurredin various clinical studies.
You can build a machinelearning model to predict
the probability of adverseevents of studies that the
model hasn't previously seen.
But there are limitshere to this in terms of
only being able to answerthis specific question.
It needs very specific data.
but what's been really impactfulmore recently is generative ai.
(06:25):
So this is where.
The model can generate newoutputs, in the form of video,
audio, or language like we seewith large language models, you
know, also referred to as LLMs.
And it's these sorts ofapproaches that are made
possible by the work that wasdone on transformer models.
and so, you know, there wasa landmark paper from 2017.
(06:48):
A year later we saw the firstrelease of the generalized
pre-trained transformer or GPTmodel, and then later iterations
of this is what were launchedin 2022 in the form of ChatGPT
that was, you know, took thepublic by storm on the internet.
And so LLMs like those, poweringthese chat bots like ChatGPT.
(07:08):
What they're doing is theymodel language as a sequence
of word fragments or, ortokens, and generated token
by token, it's this new text.
That's coming out.
It's based on all ofthe preceding text that
has gone into the model.
And so with enough training,these statistical dependencies
among these tokens, thisinterconnectedness, it
(07:30):
proves sufficient toactually produce what we
see as conversational text.
And so, you know, often it'sbasically indistinguishable from
that of a human counterpart.
the link here for something likeclinical development prediction
is that in the same way thatlanguage model learns from,
you know, the grammar, thecontextual logic of language
(07:50):
from vast bodies of, basicallyinternet scale text data sets,
clinical development modelsbased on these same types
of approaches can begin toinfer patterns of progression.
Um, you know, whenthey're trained on
data from things like.
Preclinical readouts,clinical studies, approval
documents, just to name a few.
both types of approaches canrecognize these past events,
(08:13):
but exploit these dependencies,this interconnectedness to
predict future sequences.
So that's whether, what's thenext word in the sentence?
based on a question that's beenasked in a prompt in ChatGPT
or say the next milestone,in a clinical program.
And so when we think about theselanguage models, that there's
also another aspect within,within AI that we're hearing
(08:35):
a lot about at the moment.
and this is agentic ai.
And you know, really, really,if we think about a quick
definition here, agentic AIis talking about AI systems
that can act with agency.
And what that means is thatthey have this autonomous.
ability to analyze data,but also make decisions,
execute tasks, if that'swhat you need them to do.
(08:58):
So, unlike traditionalmachine learning or
generative AI approaches.
These agents, they'redesigned to reason, but
also to plan and actindependently on these plans.
And basically this gives thiswhole new level of automation,
and adaptability and especiallywithin clinical research.
So unlike traditionallanguage-based, applications,
(09:20):
I love that I'm talking abouttraditional language models as
if this has been going on for.
Decades.
Um, and you can see howfast this is moving.
So now you can dynamicallychoose tooling to incorporate
maybe reasoning, adapttheir analysis based on
the situation at hand.
And really whatdifferentiates, agents
from, from more conventionalAI approaches is, is.
(09:43):
Really, it's a combinationof four things.
They've got an an underlyingmodel that they're utilizing.
So they might have alanguage model, they might
have a reasoning model,essentially something that
serves as like that brain.
They also have tools.
AI agents have things likemaybe, um, databases or, or
APIs, other tooling, maybeeven conventional machine
(10:05):
learning models that theycan use, to inform, the
insights they're bringing back.
But they also have, um,they have memory, they
have additional informationthat's been brought in.
And the key thing is,logic that helps the agent
figure out what do they donext based on that current
state and these decisions.
So really to, to finish thatup, the analogy to think of
(10:26):
with agentic approaches is,A language model can, if you
ask it to recommend the bestplaces to visit in Tokyo,
it will do that for you.
And it do, do it pretty well.
an agent can do that as well.
But it can also book yourflights, your accommodation,
make dinner reservations, youknow, book you into a show.
these are the sort of autonomousaspects that we're then seeing.
(10:47):
And you know, as you canimagine, within life sciences,
healthcare, clinical trials,there are, there are huge
amounts of guardrails and otherapproaches that we need to
make sure are in place beforewe just, you know, let these
agents go off autonomouslyand do what they want.
Selina Koch (11:01):
that was super interesting.
so when it comes to the scopeof agents, I guess, do you
think it's best when an agentis, designed to have a very
specific function that itfills and then you string
together different agents?
With some sort of orchestratoror how are people thinking
about building morecomplex end-to-end systems?
Greg Lever (11:23):
Yeah, no, that's exactly right.
The the real key advantage thatwe see with agentic frameworks
is that you go beyond this.
Initial situation you might haveseen, say five, 10 years ago,
where to my example, before I'vebuilt a machine learning model
that can predict adverse events.
I have this very specificquestion and then that's it.
(11:44):
I can't go beyond that.
Whereas if I, build myvery specialized, framework
of agents, some which mayhave access to data, about
adverse events, some mayhave data on, um, preclinical
readouts approval documents.
But they also havethose language models
so that yes, they canbring those insights in.
And just like you said,Selina, there's, there'll
(12:05):
be some orchestrator agent,which not only, brings
in the insights from thedifferent agents, but also.
Fully understands the intent ofthe question from the user so
that now I'm at a state where,yes, I can predict if there's
gonna be an adverse event inan upcoming clinical study,
but actually I can go beyondthat and answer additional
questions that we didn't haveto bake in at the start of
(12:28):
designing this whole framework.
And that's, that's kind ofsome of the magic and the
power that comes out of thesethings and, and actually how we
utilize this in the best way.
That's still not acompletely solved problem.
That's the excitingpiece about, agentic
frameworks at the moment.
Jeff Cranmer (12:43):
Let's bring that to target discovery.
What can and can it doin target discovery?
Greg Lever (12:48):
No, I think, I think that's really interesting.
It's, where these sourceof approaches, you know,
they're currently beingapplied, in, target discovery
in that earlier, pieceof the discovery phase.
especially whenyou're looking for.
Opportunities for selecting oridentifying, um, novel targets.
You know, we know thatthose existing pain points
(13:08):
are, how do we, how do weidentify those novel targets?
We've got all this fragmentedand disconnected data.
We, we might have limitedresources for actually doing a
really deep landscape analysis.
And actually there's just a hugerisk of missing opportunities,
and especially maybe in rareor emerging indications.
Um, and so there's acouple of approaches that
(13:29):
IQVIA uses to supportthese sorts of challenges.
And so really you can thinkabout it, is taking a language
model that's been trained onan internet scale dataset.
you can think about this as alibrary and you can think of a
standard language model like.
Someone who's read thisentire library, they can
recall what they've memorizedduring the time they read it.
(13:49):
If they're asked a question,but it's based on what
they remember, that mightbe a little bit incomplete
or it might be outdated.
And so the first step in goingbeyond this is, is known as
what's called, um, retrieval,augmented Generation or RAG.
Basically, you can imaginegiving person like a librarian
to point out here are thereally useful parts of the
library that are relevant forthe question that you have.
(14:10):
and, and maybe there mightbe updated materials that
the librarian can bring inthat the person didn't have
originally available to them.
And then the second step isto go beyond this, um, using
an approach called GraphRAG.
Basically what this does isit creates a knowledge graph
so that in practice, um, youknow, now imagine the librarian
(14:30):
is even more sophisticated.
They can, help the personunderstand these are the
important sections of thelibrary, but these are
the most relevant books.
And also these are therelationships between
concepts across these books.
Like for example, how, youknow, ideas in one chapter might
influence another, how differentauthors discuss the same
point from different angles.
(14:50):
it allows your languagemodels generate not only,
um, informative answers,but ones that are, you know,
contextually rich and connected.
And so the agentic piece,although also comes in when
we use these specializedagents to extract data
from different places.
It might be scientificliterature, trial registries,
approval documents toname a, name, a few.
(15:13):
We identify theseunderlying entities within
and across data sets.
And essentially what it,what it allows you to do is
have this optimized languagemodel where now you can start
answering questions like, whatare the future trends expected
in a particular disease area?
What are the potential growthareas to consider, within my TA
of interest, you know, are theresome indications that show.
(15:35):
more promise of theirpathologies or their
mechanisms of action.
So really, it's the sort ofthing that may enable a biotech
to move beyond just, you know,incremental innovation, and
actually pursue some, sometruly, novel approaches.
Selina Koch (15:48):
So on the target discovery front, when it
comes to ingesting all ofthose different kinds of data,
making sense of it, like whatwould be some shorter term
benchmarks of success thatyou would be looking for?
Greg Lever (16:00):
And I think this has to really.
Align with your metricsfor success in terms of
your overall R&D pipeline.
And so, if that's, if there's,if there's maybe kind of
more, more assets that you'vegot in mind and it's, it's
always gonna be these longertimescale, considerations of.
Right.
This is what I'm looking atto say de-risk my R&D pipeline
(16:24):
for the next say, 10 years.
If I'm looking at a 2035roadmap, these are the
mechanisms of action or theseare the interesting, kind of
pathologies or indications thatI need to be identifying now.
Potentially there's alsoopportunity to course correct,
um, existing R&D pipelines.
you might need to optimize insome way, but I think that's
(16:46):
where you might think abouta different type of approach.
maybe if you're thinkingabout probability of technical
and regulatory success.
Jeff Cranmer (16:54):
That's fascinating stuff, Greg.
I'm curious what, what's next?
Greg Lever (16:58):
Exactly, so, so while this, this knowledge
graph creation work is thesort of thing that's happening
right now as agentic frameworksmature and we begin to
expand on their capabilities,you can imagine agents
completing tasks like, okay.
Go through this knowledgegraph that's available to you.
Identify indicationswhere an asset may not
(17:20):
have been successful.
Based on data available.
You might be able to utilizeexisting subpopulation
analysis models to predictwhere patient subtypes may
in fact respond well to apreviously failed asset.
And you know, this, thissupports things like
defining eligibilitycriteria for future studies.
In a similar way to lookingat, you know, approved
(17:41):
therapies and identifyingpotential areas of expansion.
And this is something thatdrug repurposing as a concept
has been working on for,for a while, but it's really
going to be accelerated byagents being able to work
autonomously over this data.
But also being creativeabout how to bring
back these insights.
And I think that's somethingthat often gets underestimated
(18:03):
is the language modelscapabilities for, for
creativity, but also beingmindful that that doesn't turn
into hallucinations, which
Selina Koch (18:12):
That was what I was just gonna ask you
about how, how do you bemindful that it doesn't
turn into hallucinations?
I mean, you wanna usepositive controls.
I assume, like anyexperiment does it tell me
the things I expect to see,but are there other tips?
Greg Lever (18:24):
So I think really one of, one of the things
that's, that's crucial tothinking about is in the
same way that you have toutilize various metrics and key
performance indicators, whenyou think about any kind of,
predictive model, whether you'vejust put something together
in Excel, or you've got asimple machine learning model.
And agents have exactlythis as well, and so large
(18:44):
language models will havethings like temperature and
other parameters and aspectsthat you can look at to test.
Whether it's hallucinatingor whether it is
genuinely being creative.
And I think one example tothink about is you can think
of this future where, youmight have a, a biotech with
an interesting asset or aparticular mechanism of action.
(19:05):
There might actually be aportfolio of, say, rare diseases
with a shared mechanism,which can be impactful over
multiple, relatively smallerpatient populations and
thereby making the assetmore attractive to investors.
And it's the sort of thing that.
Right now is avery manual effort.
We'll take a lot of researchand a lot of, a lot of work
(19:25):
to generate this sort ofhypothesis, but it's the sort
of aspect that if you have anagent that has this key task,
but it's been set off to do, um,it can make it, it can make it
much, much more easier to bringback in those sorts of insights.
Selina Koch (19:39):
That's a very optimistic example given the,
uh, barriers to the businesscase in rare diseases.
so I like that one.
Okay.
Well, we've talked someabout, the applications, you
know, near and short termin clinical development and
those, for target discovery.
Well sandwiched inbetween those two things.
There's designing a molecule,predicting whether or not
it's going to be successful.
(20:00):
let's dig intothat a little bit.
Greg Lever (20:02):
Absolutely.
So I think one of the thingsto think about, in terms
of some of the forwardlooking future thinking
aspects as well, is that.
Once a, a target has beenidentified, the sorts of
things that maybe earlystage biotechs can think of
is that AI agents are goingto increasingly be used to
assess things like technical,regulatory, operational
(20:25):
success, and really how theycan reduce their risk and
increase confidence in their,in their development strategy.
And so the way, the way thesesorts of things will work is
there'll be these specializedagents that will be able to
understand aspects like, okay,I might have a PTRS agent,
which is looking at mechanismsof action, patient populations
(20:46):
related, existing approvals.
I may also have dataretrieval agents looking
across scientific literatureor conference abstracts.
I may have trial search agentslooking across existing clinical
trials, and so this then getsall brought back together by
an orchestrator agent that notonly brings in the insights from
(21:06):
these specialized agents, but.
Can really understand theintent of my user's question,
which is basically sayingwhat's the potential of
my early stage, asset?
And then by extension we canutilize existing molecular
design approaches or targetdesign approaches to say, and
is there some better designthat I could think of here, um,
(21:28):
that will like really acceleratemy clinical development?
Selina Koch (21:31):
So in protein design seems to be a
little further ahead ofsay small molecule design,
unless I've misunderstood.
Um, but there.
This neat idea I heard recentlyof, so if language models,
they tokenize language, asyou were saying earlier, where
the smallest unit of meaningis the word as opposed to
(21:54):
the way we learn language isbuilding it up from letters.
that there is an analogy inthe small molecule space of
tokens or bits of moleculeswith certain functions and that
you might be able to have like.
A tokenized library offunctional components that
could be built into anyway.
What, tell U.S. a littlebit on the small molecule
(22:15):
front, what you're hearing.
That could be abe a step change?
Greg Lever (22:19):
So I think what we can think of in terms
of the protein and antibodymodeling space is things like
protein language models wherethat fundamental token or
fragment is either an aminoacid, if we're thinking about
proteins or nucleic acids.
If we're thinking aboutother structures like
DNA and otherwise.
And I think really.
(22:40):
You can think about molecularlanguage models that actually
build up this concept.
at the atomic level, andpotentially also at the,
the electron level as well.
And that's actually where alot of my, academic, studies
and academic research was in,is in how do we build up a
sufficient electron densityto understand how these small
(23:03):
molecules can best interact.
Um, with their targets and beingimpactful in a clinical setting.
And it's something that thelanguage models are definitely
going to, to really accelerate,um, in the coming years.
And it's a, it's a space that wereally need to keep an eye on.
Selina Koch (23:17):
That was very cool.
And then if we wannaeven go further afield, I
guess in our imaginationshere, what can happen?
I just heard a reallyinteresting talk at our Grand
Rounds conference on, um,the possibility of quantum
computing, really speedingup simulations, particularly
around all the electronicstates and things like you
were just talking about.
(23:38):
but that's, not here yet.
Greg Lever (23:40):
And that's probably also an entire podcast
Selina Koch (23:42):
Yeah.
Yeah.
Yeah.
Lauren Martz (23:44):
I think it would be great if we could
get into a little bit aboutthe clinical trial design
AI applications, where weare now, and where we could
be, you know, in the future?
Greg Lever (23:55):
Yeah, that, that's a really interesting
point and I think there's alot to to speak about, but
I think what I would say is.
AI is really transformingtrial design.
It's, it's doing thingslike quantifying site
and patient burden.
being able to look at thingslike protocol complexity,
but also simulatingoperational outcomes.
So, not only can we use theseadvanced analytics and other
(24:18):
conventional machine learningapproaches to really predict how
different decisions within thedesign process, like eligibility
criteria and visit schedules.
How they might impactrecruitment rates, um,
patient retention, um,or site performance.
But now with, increasinglygenerative ai, it's
enabling these sponsors to,to really optimize these
(24:40):
protocols, really say.
Okay, this is what may happenin the current state, but how
do I make this even better?
How do I forecastthat site activation?
How do I select sites with thehighest likelihood of success?
And so really it's, it'sgetting to those pain points
of, we know there's a lotof protocol complexity.
This is slowing recruitment.
We know there's uncertaintyin these operational outcomes.
(25:04):
and also we know thatthe biotechs have these
resource constraints.
And so really what youcan do is you can, you can
begin to demonstrate thisvalue and sort of leapfrog
across a lot of theseconstraints, by utilizing ai.
Selina Koch (25:16):
So looking across these different
domains, target discovery,molecule discovery, clinical
trial prediction design.
When you're advising biotechstoday, or if you were to, um,
and you had to give just a veryshort list of like, these are
the most useful applicationsright now across those domains,
like what would you say.
Greg Lever (25:34):
I think it's about making sure that you've, you've
tried to have that coverageacross the clinical development
life cycle or the, or the piecesthat are important for you.
to make sure that, okay,in the discovery phase,
maybe you are thinking moreabout target id, generative
approaches for design.
when you do get that, assetinto the clinic, really how are
(25:57):
you optimizing, your protocol?
And then also maybe whenyou are thinking, about
the regulatory space.
how are you utilizing AIto really think about what
are the compliance changesthat are happening globally
or in my target market?
And how do I need to basicallyde-risk, my development and
be able to anticipate thesethings and act earlier?
(26:17):
And that's, that's in a nutshellwhat these AI approaches are
providing you the capability of.
Selina Koch (26:23):
Is there any application right now where
you'd say, just steer clearof that one, or don't expect
a lot from that just yet?
Greg Lever (26:30):
So I think one of the biggest challenges that we
might want to leave to slightlymore, well-funded, outfits
is the notion of in silicoclinical trial simulation,
and this is an aspect that hashad lots of attention over.
Many years from manydifferent types of approaches.
It's something that, generativeai, agentic ai, and specifically
(26:54):
foundation models are goingto have a massive impact in.
But it's still very early days.
Jeff Cranmer (27:00):
well, we've been talking AI in biotech
with Greg Lever of IQVIAand Greg, I'd just like to
get your closing thoughts?
Greg Lever (27:10):
Yeah, absolutely.
I think really one ofthe things to think
about is that for those.
Biotechs, early stage biotechs,those sort of outfits.
agentic AI really offers theability to compete at scale.
and whether it's identifyingnovel opportunities, designing
(27:31):
smarter trials, you know,accelerating development
with, with fewer resources,Not only that, the other
approaches that we talkedabout, be it generative ai
and also conventional machinelearning, could be helpful in,
in attracting investment, youknow, utilizing these approaches
to give this more objective,data-driven view of asset
(27:52):
risk, but also asset value.
And so supporting thingslike capital allocation,
the potential to de-riskthese investment decisions.
I think that's gonnabe really impactful.
Jeff Cranmer (28:02):
Excellent.
Well, Greg, thank youso much for joining us.
the way things are moving soquickly, we're gonna have to
have you, uh, back on tomorrow.
I don't, I don't know.
It's, uh, tough to keepup with everything that's
going on in, uh, how AIis helping to improve,
how we work in biotech.
so once again, this hasbeen, uh, the BioCentury
analyst team, speaking withGreg Lever, Director of AI
(28:24):
Solutions Delivery from IQVIA.
a special thanks to IQVIABiotech our sponsor, as well
as Kendall Square Orchestra,the Boston based, ensemble
that, does the music for all ofBioCentury's podcast Tickets on
sale now for their new season.
(28:44):
we will catch you onMonday with the regular
BioCentury This Week podcast.
Alanna (28:51):
BioCentury would like to thank IQVIA
Biotech for supporting theBioCentury This Week podcast.
To learn more about how IQVIABiotech can help you turn your
vision into venture capital, goto IQVIABiotech.com/visionaries
[AI-generated transcript.]
Eric Pierce (00:02):
BioCentury This Week is brought to
you by IQVIA Biotech.
For biotech companies strivingto bring innovative therapies
to market and maximize patientimpact, IQVIA Biotech is
the trusted CRO of choice.
Backed by 25 years ofunparalleled experience
and deep therapeuticexpertise, our full-service
clinical developmentsolutions are purpose-built
(00:23):
to accelerate success.
IQVIA Biotech helpsearly-stage biotechs de-risk
by developing strategicclinical development plans,
guiding drug candidates alongthe most promising pathways.
Leverage data-driven modelsand dynamic tools to craft
a compelling value story andmaintain momentum through every
phase of drug development.
Jeff Cranmer (00:49):
AI is the buzzword everywhere these days and
biotech is no stranger to that.
Welcome to a specialedition of the BioCentury
This Week podcast.
Today I'm very pleased.
we have a special guest fromour sponsor, Greg Lever,
he's the Director of AISolutions Delivery at IQVIA.
(01:13):
And joining me as wellare my colleagues and
podcast, regulars, SelinaKoch and Lauren Martz.
And we're gonna cut rightto the chase, Selina, when
you think of AI in biotech,uh, what does it say to you?
Selina Koch (01:29):
right now if I try to take a step back and
like take in all these changesthat are sweeping through the
industry so rapidly, it kind ofmakes my head spin a little bit.
conversations about, howto use AI and agentic AI and
drug development, are now,you know, being happening
in, so it seems like in everysingle company, at every
(01:50):
investment firm, and it's even.
Maybe even hard to go toa meeting where the topic
doesn't come up right?
So things are changing atkind of dizzying speed, but
because of that, we're in thislike moment where there's this
wide variability across theindustry in terms of how much
or little people understandabout the technology and
how proactive they are inengaging with it, I think.
(02:13):
And so I think it'sa really nice time to
have this conversationto kind of take stock.
Of where agentic AI is,making some practical inroads
and, where it has, otherapplications in the future.
if I could say one morething, I don't think that's
always obvious, like where AIis gonna make contribution.
like I don't think manypeople would've guessed
that writing code would beamong the first breakthrough
(02:35):
applications of AI that.
First jobs under threat would beprogramming jobs, for example.
But you know, in retrospectit makes sense when you think
about the data sets thatare available in that space.
There are, well, I don'tknow, or Greg might know
this, um, tens of millionsof programmers who regularly
check code into GitHub.
So there's this really bigrepository that not only has
(02:58):
a lot of code, but has like acode progression, leading to
code working or not working.
So you have.
Also this referenceto ground truth.
but when I think aboutbiology, it's really not
that straightforward, right?
We have this like fragmentedevidence ecosystem.
We have a few large scale datasets would have that kind of
reference to ground truth.
(03:18):
that makes me question in thenear term where there's gonna
be practical application.
Like if you think ofprotein design, for example.
I can imagine that optimizationof antibodies and proteins
along certain parameterswill, will be happening.
It is happening, right?
But wholesale de novo designcould take a while longer.
I dunno.
If you all disagree,feel free to chime in.
(03:39):
But
Jeff Cranmer (03:40):
Lauren, do you have any thoughts
that you wanna throw outthere before we, uh, go?
I see Greg nodding alongand we'll, we're gonna
bring him in in just aminute, but, uh, Lauren.
Lauren Martz (03:48):
Yeah, I think we can pass it to Greg momentarily,
just to follow up on some ofthe ways that this technology
might be finding its way intoour industry and already is.
there's protein design,there's target discovery.
And that goes all the waythrough to ways that, AI can
support clinical trial design.
you know, we've been hearingabout this for years.
(04:10):
I think a lot of what we'veheard initially is that.
This can be used to help withsite selection, you know,
with, finding the locationsand all the different
dimensions that go into that.
This is a place where datacan have a big impact.
You know, where the patientsare, where the right
investigators are, um, youknow, how many trials are being
run in that, in that site.
This is something that we,we've already seen this in
action, from Greg, I'd loveto hear more about how AI
(04:34):
can support clinical trialdesign and everything in this
spectrum of drug discovery,through drug development.
Jeff Cranmer (04:41):
Alright, well let's bring in Greg.
It's Greg Lever, director ofAI Solutions Delivery at IQVIA.
And Greg, I know, we'vechatted a bit before and
I think you thought it wasimportant to kind of set some
definitions to get us oriented.
why don't I hand it over to you?
Greg Lever (04:59):
Thank you.
Yeah, thanks for having me.
It's great to be here.
Yeah, and I think to Selina'spoint, there's a lot going on
in the industry and I thinkit can be quite difficult
to get a, a proper grip on alot of the terminology that's
out there and the differentapproaches and, and also why
they matter and, and, andwhat it is that we can, we can
actually, you know, use here.
So I think to act as a, astarting point, it can be,
(05:21):
helpful to think about.
How all the different termsinterconnect is specifically
aspects like ai, machinelearning, generative ai,
ai, and if we start with ai,as a broad term, artificial
intelligence is really, youknow, used to describe all of
these sorts of initiatives.
And it was, started asan academic discipline
(05:41):
in the mid 1950s.
and there are now vastdifferent types of
approaches and algorithmswithin this space and.
One of these approachesis, machine learning.
So this is where, you know,typically you have algorithms
that can be used to train amodel so that it can learn from
data, make predictions basedon unseen, but related data.
So, as a more concreteexample, say I have data on
(06:05):
adverse events that occurredin various clinical studies.
You can build a machinelearning model to predict
the probability of adverseevents of studies that the
model hasn't previously seen.
But there are limitshere to this in terms of
only being able to answerthis specific question.
It needs very specific data.
but what's been really impactfulmore recently is generative ai.
(06:25):
So this is where.
The model can generate newoutputs, in the form of video,
audio, or language like we seewith large language models, you
know, also referred to as LLMs.
And it's these sorts ofapproaches that are made
possible by the work that wasdone on transformer models.
and so, you know, there wasa landmark paper from 2017.
(06:48):
A year later we saw the firstrelease of the generalized
pre-trained transformer or GPTmodel, and then later iterations
of this is what were launchedin 2022 in the form of ChatGPT
that was, you know, took thepublic by storm on the internet.
And so LLMs like those, poweringthese chat bots like ChatGPT.
(07:08):
What they're doing is theymodel language as a sequence
of word fragments or, ortokens, and generated token
by token, it's this new text.
That's coming out.
It's based on all ofthe preceding text that
has gone into the model.
And so with enough training,these statistical dependencies
among these tokens, thisinterconnectedness, it
(07:30):
proves sufficient toactually produce what we
see as conversational text.
And so, you know, often it'sbasically indistinguishable from
that of a human counterpart.
the link here for something likeclinical development prediction
is that in the same way thatlanguage model learns from,
you know, the grammar, thecontextual logic of language
(07:50):
from vast bodies of, basicallyinternet scale text data sets,
clinical development modelsbased on these same types
of approaches can begin toinfer patterns of progression.
Um, you know, whenthey're trained on
data from things like.
Preclinical readouts,clinical studies, approval
documents, just to name a few.
both types of approaches canrecognize these past events,
(08:13):
but exploit these dependencies,this interconnectedness to
predict future sequences.
So that's whether, what's thenext word in the sentence?
based on a question that's beenasked in a prompt in ChatGPT
or say the next milestone,in a clinical program.
And so when we think about theselanguage models, that there's
also another aspect within,within AI that we're hearing
(08:35):
a lot about at the moment.
and this is agentic ai.
And you know, really, really,if we think about a quick
definition here, agentic AIis talking about AI systems
that can act with agency.
And what that means is thatthey have this autonomous.
ability to analyze data,but also make decisions,
execute tasks, if that'swhat you need them to do.
(08:58):
So, unlike traditionalmachine learning or
generative AI approaches.
These agents, they'redesigned to reason, but
also to plan and actindependently on these plans.
And basically this gives thiswhole new level of automation,
and adaptability and especiallywithin clinical research.
So unlike traditionallanguage-based, applications,
(09:20):
I love that I'm talking abouttraditional language models as
if this has been going on for.
Decades.
Um, and you can see howfast this is moving.
So now you can dynamicallychoose tooling to incorporate
maybe reasoning, adapttheir analysis based on
the situation at hand.
And really whatdifferentiates, agents
from, from more conventionalAI approaches is, is.
(09:43):
Really, it's a combinationof four things.
They've got an an underlyingmodel that they're utilizing.
So they might have alanguage model, they might
have a reasoning model,essentially something that
serves as like that brain.
They also have tools.
AI agents have things likemaybe, um, databases or, or
APIs, other tooling, maybeeven conventional machine
(10:05):
learning models that theycan use, to inform, the
insights they're bringing back.
But they also have, um,they have memory, they
have additional informationthat's been brought in.
And the key thing is,logic that helps the agent
figure out what do they donext based on that current
state and these decisions.
So really to, to finish thatup, the analogy to think of
(10:26):
with agentic approaches is,A language model can, if you
ask it to recommend the bestplaces to visit in Tokyo,
it will do that for you.
And it do, do it pretty well.
an agent can do that as well.
But it can also book yourflights, your accommodation,
make dinner reservations, youknow, book you into a show.
these are the sort of autonomousaspects that we're then seeing.
(10:47):
And you know, as you canimagine, within life sciences,
healthcare, clinical trials,there are, there are huge
amounts of guardrails and otherapproaches that we need to
make sure are in place beforewe just, you know, let these
agents go off autonomouslyand do what they want.
Selina Koch (11:01):
that was super interesting.
so when it comes to the scopeof agents, I guess, do you
think it's best when an agentis, designed to have a very
specific function that itfills and then you string
together different agents?
With some sort of orchestratoror how are people thinking
about building morecomplex end-to-end systems?
Greg Lever (11:23):
Yeah, no, that's exactly right.
The the real key advantage thatwe see with agentic frameworks
is that you go beyond this.
Initial situation you might haveseen, say five, 10 years ago,
where to my example, before I'vebuilt a machine learning model
that can predict adverse events.
I have this very specificquestion and then that's it.
(11:44):
I can't go beyond that.
Whereas if I, build myvery specialized, framework
of agents, some which mayhave access to data, about
adverse events, some mayhave data on, um, preclinical
readouts approval documents.
But they also havethose language models
so that yes, they canbring those insights in.
And just like you said,Selina, there's, there'll
(12:05):
be some orchestrator agent,which not only, brings
in the insights from thedifferent agents, but also.
Fully understands the intent ofthe question from the user so
that now I'm at a state where,yes, I can predict if there's
gonna be an adverse event inan upcoming clinical study,
but actually I can go beyondthat and answer additional
questions that we didn't haveto bake in at the start of
(12:28):
designing this whole framework.
And that's, that's kind ofsome of the magic and the
power that comes out of thesethings and, and actually how we
utilize this in the best way.
That's still not acompletely solved problem.
That's the excitingpiece about, agentic
frameworks at the moment.
Jeff Cranmer (12:43):
Let's bring that to target discovery.
What can and can it doin target discovery?
Greg Lever (12:48):
No, I think, I think that's really interesting.
It's, where these sourceof approaches, you know,
they're currently beingapplied, in, target discovery
in that earlier, pieceof the discovery phase.
especially whenyou're looking for.
Opportunities for selecting oridentifying, um, novel targets.
You know, we know thatthose existing pain points
(13:08):
are, how do we, how do weidentify those novel targets?
We've got all this fragmentedand disconnected data.
We, we might have limitedresources for actually doing a
really deep landscape analysis.
And actually there's just a hugerisk of missing opportunities,
and especially maybe in rareor emerging indications.
Um, and so there's acouple of approaches that
(13:29):
IQVIA uses to supportthese sorts of challenges.
And so really you can thinkabout it, is taking a language
model that's been trained onan internet scale dataset.
you can think about this as alibrary and you can think of a
standard language model like.
Someone who's read thisentire library, they can
recall what they've memorizedduring the time they read it.
(13:49):
If they're asked a question,but it's based on what
they remember, that mightbe a little bit incomplete
or it might be outdated.
And so the first step in goingbeyond this is, is known as
what's called, um, retrieval,augmented Generation or RAG.
Basically, you can imaginegiving person like a librarian
to point out here are thereally useful parts of the
library that are relevant forthe question that you have.
(14:10):
and, and maybe there mightbe updated materials that
the librarian can bring inthat the person didn't have
originally available to them.
And then the second step isto go beyond this, um, using
an approach called GraphRAG.
Basically what this does isit creates a knowledge graph
so that in practice, um, youknow, now imagine the librarian
(14:30):
is even more sophisticated.
They can, help the personunderstand these are the
important sections of thelibrary, but these are
the most relevant books.
And also these are therelationships between
concepts across these books.
Like for example, how, youknow, ideas in one chapter might
influence another, how differentauthors discuss the same
point from different angles.
(14:50):
it allows your languagemodels generate not only,
um, informative answers,but ones that are, you know,
contextually rich and connected.
And so the agentic piece,although also comes in when
we use these specializedagents to extract data
from different places.
It might be scientificliterature, trial registries,
approval documents toname a, name, a few.
(15:13):
We identify theseunderlying entities within
and across data sets.
And essentially what it,what it allows you to do is
have this optimized languagemodel where now you can start
answering questions like, whatare the future trends expected
in a particular disease area?
What are the potential growthareas to consider, within my TA
of interest, you know, are theresome indications that show.
(15:35):
more promise of theirpathologies or their
mechanisms of action.
So really, it's the sort ofthing that may enable a biotech
to move beyond just, you know,incremental innovation, and
actually pursue some, sometruly, novel approaches.
Selina Koch (15:48):
So on the target discovery front, when it
comes to ingesting all ofthose different kinds of data,
making sense of it, like whatwould be some shorter term
benchmarks of success thatyou would be looking for?
Greg Lever (16:00):
And I think this has to really.
Align with your metricsfor success in terms of
your overall R&D pipeline.
And so, if that's, if there's,if there's maybe kind of
more, more assets that you'vegot in mind and it's, it's
always gonna be these longertimescale, considerations of.
Right.
This is what I'm looking atto say de-risk my R&D pipeline
(16:24):
for the next say, 10 years.
If I'm looking at a 2035roadmap, these are the
mechanisms of action or theseare the interesting, kind of
pathologies or indications thatI need to be identifying now.
Potentially there's alsoopportunity to course correct,
um, existing R&D pipelines.
you might need to optimize insome way, but I think that's
(16:46):
where you might think abouta different type of approach.
maybe if you're thinkingabout probability of technical
and regulatory success.
Jeff Cranmer (16:54):
That's fascinating stuff, Greg.
I'm curious what, what's next?
Greg Lever (16:58):
Exactly, so, so while this, this knowledge
graph creation work is thesort of thing that's happening
right now as agentic frameworksmature and we begin to
expand on their capabilities,you can imagine agents
completing tasks like, okay.
Go through this knowledgegraph that's available to you.
Identify indicationswhere an asset may not
(17:20):
have been successful.
Based on data available.
You might be able to utilizeexisting subpopulation
analysis models to predictwhere patient subtypes may
in fact respond well to apreviously failed asset.
And you know, this, thissupports things like
defining eligibilitycriteria for future studies.
In a similar way to lookingat, you know, approved
(17:41):
therapies and identifyingpotential areas of expansion.
And this is something thatdrug repurposing as a concept
has been working on for,for a while, but it's really
going to be accelerated byagents being able to work
autonomously over this data.
But also being creativeabout how to bring
back these insights.
And I think that's somethingthat often gets underestimated
(18:03):
is the language modelscapabilities for, for
creativity, but also beingmindful that that doesn't turn
into hallucinations, which
Selina Koch (18:12):
That was what I was just gonna ask you
about how, how do you bemindful that it doesn't
turn into hallucinations?
I mean, you wanna usepositive controls.
I assume, like anyexperiment does it tell me
the things I expect to see,but are there other tips?
Greg Lever (18:24):
So I think really one of, one of the things
that's, that's crucial tothinking about is in the
same way that you have toutilize various metrics and key
performance indicators, whenyou think about any kind of,
predictive model, whether you'vejust put something together
in Excel, or you've got asimple machine learning model.
And agents have exactlythis as well, and so large
(18:44):
language models will havethings like temperature and
other parameters and aspectsthat you can look at to test.
Whether it's hallucinatingor whether it is
genuinely being creative.
And I think one example tothink about is you can think
of this future where, youmight have a, a biotech with
an interesting asset or aparticular mechanism of action.
(19:05):
There might actually be aportfolio of, say, rare diseases
with a shared mechanism,which can be impactful over
multiple, relatively smallerpatient populations and
thereby making the assetmore attractive to investors.
And it's the sort of thing that.
Right now is avery manual effort.
We'll take a lot of researchand a lot of, a lot of work
(19:25):
to generate this sort ofhypothesis, but it's the sort
of aspect that if you have anagent that has this key task,
but it's been set off to do, um,it can make it, it can make it
much, much more easier to bringback in those sorts of insights.
Selina Koch (19:39):
That's a very optimistic example given the,
uh, barriers to the businesscase in rare diseases.
so I like that one.
Okay.
Well, we've talked someabout, the applications, you
know, near and short termin clinical development and
those, for target discovery.
Well sandwiched inbetween those two things.
There's designing a molecule,predicting whether or not
it's going to be successful.
(20:00):
let's dig intothat a little bit.
Greg Lever (20:02):
Absolutely.
So I think one of the thingsto think about, in terms
of some of the forwardlooking future thinking
aspects as well, is that.
Once a, a target has beenidentified, the sorts of
things that maybe earlystage biotechs can think of
is that AI agents are goingto increasingly be used to
assess things like technical,regulatory, operational
(20:25):
success, and really how theycan reduce their risk and
increase confidence in their,in their development strategy.
And so the way, the way thesesorts of things will work is
there'll be these specializedagents that will be able to
understand aspects like, okay,I might have a PTRS agent,
which is looking at mechanismsof action, patient populations
(20:46):
related, existing approvals.
I may also have dataretrieval agents looking
across scientific literatureor conference abstracts.
I may have trial search agentslooking across existing clinical
trials, and so this then getsall brought back together by
an orchestrator agent that notonly brings in the insights from
(21:06):
these specialized agents, but.
Can really understand theintent of my user's question,
which is basically sayingwhat's the potential of
my early stage, asset?
And then by extension we canutilize existing molecular
design approaches or targetdesign approaches to say, and
is there some better designthat I could think of here, um,
(21:28):
that will like really acceleratemy clinical development?
Selina Koch (21:31):
So in protein design seems to be a
little further ahead ofsay small molecule design,
unless I've misunderstood.
Um, but there.
This neat idea I heard recentlyof, so if language models,
they tokenize language, asyou were saying earlier, where
the smallest unit of meaningis the word as opposed to
(21:54):
the way we learn language isbuilding it up from letters.
that there is an analogy inthe small molecule space of
tokens or bits of moleculeswith certain functions and that
you might be able to have like.
A tokenized library offunctional components that
could be built into anyway.
What, tell U.S. a littlebit on the small molecule
(22:15):
front, what you're hearing.
That could be abe a step change?
Greg Lever (22:19):
So I think what we can think of in terms
of the protein and antibodymodeling space is things like
protein language models wherethat fundamental token or
fragment is either an aminoacid, if we're thinking about
proteins or nucleic acids.
If we're thinking aboutother structures like
DNA and otherwise.
And I think really.
(22:40):
You can think about molecularlanguage models that actually
build up this concept.
at the atomic level, andpotentially also at the,
the electron level as well.
And that's actually where alot of my, academic, studies
and academic research was in,is in how do we build up a
sufficient electron densityto understand how these small
(23:03):
molecules can best interact.
Um, with their targets and beingimpactful in a clinical setting.
And it's something that thelanguage models are definitely
going to, to really accelerate,um, in the coming years.
And it's a, it's a space that wereally need to keep an eye on.
Selina Koch (23:17):
That was very cool.
And then if we wannaeven go further afield, I
guess in our imaginationshere, what can happen?
I just heard a reallyinteresting talk at our Grand
Rounds conference on, um,the possibility of quantum
computing, really speedingup simulations, particularly
around all the electronicstates and things like you
were just talking about.
(23:38):
but that's, not here yet.
Greg Lever (23:40):
And that's probably also an entire podcast
Selina Koch (23:42):
Yeah.
Yeah.
Yeah.
Lauren Martz (23:44):
I think it would be great if we could
get into a little bit aboutthe clinical trial design
AI applications, where weare now, and where we could
be, you know, in the future?
Greg Lever (23:55):
Yeah, that, that's a really interesting
point and I think there's alot to to speak about, but
I think what I would say is.
AI is really transformingtrial design.
It's, it's doing thingslike quantifying site
and patient burden.
being able to look at thingslike protocol complexity,
but also simulatingoperational outcomes.
So, not only can we use theseadvanced analytics and other
(24:18):
conventional machine learningapproaches to really predict how
different decisions within thedesign process, like eligibility
criteria and visit schedules.
How they might impactrecruitment rates, um,
patient retention, um,or site performance.
But now with, increasinglygenerative ai, it's
enabling these sponsors to,to really optimize these
(24:40):
protocols, really say.
Okay, this is what may happenin the current state, but how
do I make this even better?
How do I forecastthat site activation?
How do I select sites with thehighest likelihood of success?
And so really it's, it'sgetting to those pain points
of, we know there's a lotof protocol complexity.
This is slowing recruitment.
We know there's uncertaintyin these operational outcomes.
(25:04):
and also we know thatthe biotechs have these
resource constraints.
And so really what youcan do is you can, you can
begin to demonstrate thisvalue and sort of leapfrog
across a lot of theseconstraints, by utilizing ai.
Selina Koch (25:16):
So looking across these different
domains, target discovery,molecule discovery, clinical
trial prediction design.
When you're advising biotechstoday, or if you were to, um,
and you had to give just a veryshort list of like, these are
the most useful applicationsright now across those domains,
like what would you say.
Greg Lever (25:34):
I think it's about making sure that you've, you've
tried to have that coverageacross the clinical development
life cycle or the, or the piecesthat are important for you.
to make sure that, okay,in the discovery phase,
maybe you are thinking moreabout target id, generative
approaches for design.
when you do get that, assetinto the clinic, really how are
(25:57):
you optimizing, your protocol?
And then also maybe whenyou are thinking, about
the regulatory space.
how are you utilizing AIto really think about what
are the compliance changesthat are happening globally
or in my target market?
And how do I need to basicallyde-risk, my development and
be able to anticipate thesethings and act earlier?
(26:17):
And that's, that's in a nutshellwhat these AI approaches are
providing you the capability of.
Selina Koch (26:23):
Is there any application right now where
you'd say, just steer clearof that one, or don't expect
a lot from that just yet?
Greg Lever (26:30):
So I think one of the biggest challenges that we
might want to leave to slightlymore, well-funded, outfits
is the notion of in silicoclinical trial simulation,
and this is an aspect that hashad lots of attention over.
Many years from manydifferent types of approaches.
It's something that, generativeai, agentic ai, and specifically
(26:54):
foundation models are goingto have a massive impact in.
But it's still very early days.
Jeff Cranmer (27:00):
well, we've been talking AI in biotech
with Greg Lever of IQVIAand Greg, I'd just like to
get your closing thoughts?
Greg Lever (27:10):
Yeah, absolutely.
I think really one ofthe things to think
about is that for those.
Biotechs, early stage biotechs,those sort of outfits.
agentic AI really offers theability to compete at scale.
and whether it's identifyingnovel opportunities, designing
(27:31):
smarter trials, you know,accelerating development
with, with fewer resources,Not only that, the other
approaches that we talkedabout, be it generative ai
and also conventional machinelearning, could be helpful in,
in attracting investment, youknow, utilizing these approaches
to give this more objective,data-driven view of asset
(27:52):
risk, but also asset value.
And so supporting thingslike capital allocation,
the potential to de-riskthese investment decisions.
I think that's gonnabe really impactful.
Jeff Cranmer (28:02):
Excellent.
Well, Greg, thank youso much for joining us.
the way things are moving soquickly, we're gonna have to
have you, uh, back on tomorrow.
I don't, I don't know.
It's, uh, tough to keepup with everything that's
going on in, uh, how AIis helping to improve,
how we work in biotech.
so once again, this hasbeen, uh, the BioCentury
analyst team, speaking withGreg Lever, Director of AI
(28:24):
Solutions Delivery from IQVIA.
a special thanks to IQVIABiotech our sponsor, as well
as Kendall Square Orchestra,the Boston based, ensemble
that, does the music for all ofBioCentury's podcast Tickets on
sale now for their new season.
(28:44):
we will catch you onMonday with the regular
BioCentury This Week podcast.
Alanna (28:51):
BioCentury would like to thank IQVIA
Biotech for supporting theBioCentury This Week podcast.
To learn more about how IQVIABiotech can help you turn your
vision into venture capital, goto IQVIABiotech.com/visionaries
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!