Thank you OK thank you for
this kind introduction again I think.

It's very nice in sort of telling you
about how an interesting career or

bizarre career I've had because
frankly going from academia to

government to industry is not
something that happens a lot because

of chance it was pure luck but
I've constantly been able or

been lucky enough to be at
the interface of multiple

multiple fields My father was
a teacher of mathematics so

in addition to what and Russell
tell you I'll tell you my personal.

Make up if you will that lead to being
able to or wanting to do this and

to share with you the lessons that
I learned so the 1st thing was my

father was a mathematician is a teacher
in mathematics and so I grew up actually

never thinking that I would be a physician
he wanted me to be an engineer or

a scientist but not a physician not
a lawyer Those are the things he said you

can do anything but
not position not a lawyer and so

so when I grew up you know we
had at home I had 6 brothers and

we were constantly challenging ourselves
with brain teasers and mathematics and and

physics problems and
all kinds of things that I loved and

I you know if my 6 brothers
5 are engineers so

you can imagine the detraction there but
then one day I was a student I

went to volunteer in the mountains of
Algeria from Algeria from North Africa so

I'm an immigrant I'm not born here and
I went to the mountains and and then we

saw the misery of people in the mountains
of North Africa about tuberculosis and

and I saw an event that really
touched me because it was a patient

she would had tuberculosis and
she went to a medical center and

they didn't have anything because this
was after the war and the guy had a you

know these old icebox where you put the
ice and that was to keep the vaccine so

he opened the icebox and he said put
your chest here she had you know for

you Do you know her fever was probably
one or 2 something like this and

she just went against the ice box and

she got a cold feeling and she felt
good and then you close that say well

that was the X. ray and she helped thank
you very much and then you give a shot.

I don't know what struck him I see maybe
and she laughs back to the I was really

touched by that I thought my god they're
better solutions that we could do so.

For whatever reason I decided going to
medical school I thought I was more human

and hearing so.

So so I did and then my father was
really upset with me and he said

you're lowering yourself medical school
is just basically rote learning you don't

learn recipes about drugs and there's
no no intelligence in medical school.

And so I did go to medical school
because you know when you're.

Young you'd like to not do what your
father wants you to do and then you're

always over or is worried that you won't
do as well as either so so I did I went to

medical school and I found out that he was
right after 34 years I was bored to death.

And it was boring and you could have
a conversation with a teacher that

meant anything from the point
of view of science and

it was all empirical observation all
there was nothing quantitative and

I always try to put whatever they
said in an equation system and

I couldn't and so I was an extremely
frustrated I was going to quit medical

school I said I made a mistake I don't
want to go to medical school and then a.

Radiologist showed me the 1st
class can image he said

look look look at this and I was a medical
student and he showed me these very

big pixel you know you barely
recognize it was a human head and

you could see the ventricles and
that's how does that work so

well you know you have an extra 2 turns
around and they reconstruct with for

you transforms and I fell in love with
that idea of combining physics and

mathematics and biology so I decided
that I was going to do imaging and

that's when I decided to come to the U.S.
to do research and

I was lucky enough to end up at Hopkins
and the rest is basically what I'm dressed

told you so
today at the end of that I wanted to.

Thank Bob Niram for
inviting me and dress to.

I'm here and asking me to speak about
OK what did you really learn and

what do you remember that we can use so so
that's the topic of this lecture today so

I think when you look at any system you
have to really take a different altitude

as you look at the system so
I call it the 100000 foot view

1st never look at something from
inside out don't look at your lab and

then what else is there look the other way
around say what is the total system and

so what you realized very quickly is that
academia industry and government truly

are an ecosystem they're
separate No question about it but

their agenda is a complimentary and if you
haven't really been around all 3 you don't

get that you don't perceive that so I'm
going to try to share with you why that is

and how you can navigate that
ecosystem both as a student or

as an institution so the 1st thing
that you have to understand is you

go down to 10000 feet now that
you're seeing the government and

all of that you say OK let
me understand each component

of the system we're engineers right so
the 1st thing you have to understand I

continue and when you look at
academia you have to say OK what

are the fundamental drivers of academia
what do we do really who are weak and

if you get that then the understand
where the connections to the the rest of

the system I would say the number one
role of academia is to be the guardian

of the intellectual heritage of mankind
right everything that was learned over

thousands of years is our heritage and
transmitting that heritage is one of

the most powerful evolutionary

force because frankly
natural evolution is.

Basically based on transmitting
information through D.N.A. mutations right

and we've done that pretty well for
millions and billions of years but

knowledge and Teligent have allowed
a different way of accelerating

evolution by being able to transfer
knowledge through education so you don't

have to have it in your genes you just go
to a great school like Georgia Tech and

you get that evolution power behind you so
that's one of the saying so

the transmission of knowledge in
the memory that we have of it is a key

fundamental aspect of academia
obviously that leads to creating

the human capital that any civilization
any society needs to be able to.

Influence its environment and
that's what I think.

I could deem is also important as
a social escalator you know anybody will

tell you as you look at tables of
compensations that if you have a high

degree you tend to be more
compensated if you don't and

if you have a strong education you tend to
do better socially than if you don't so so

these situations are very key
to the social balance and

the evolution of society but
the fundamental drive is to understand

the world so whether it be basic or
early applied research that's what we do

that's what we want to do we want to have
the young capital of society come in and

really try to break barriers and knowledge
that's what I can do is all about

that I can dream is very resilient so
here I have a 500 to a 1000 year cycle and

the question of 500000 years came from
a study that was done by a gentleman named

Norman Augustine and he looked at the
history of governments and I could and.

And industry and said What's
the lifetime the average lifetime

of Fame institution and I get them
against edition as opposed to another and

you find out that if you look at the 100
universities that were existing in 1500

and you look today 75 percent of
them are still alive and well and

you can see that the lifetime of
universities in the 500000 years range

Oxford 1200 and so on so
they have a long cycle time

very very long prospects not get
the what about government so

if you look at government you have to also
say well OK so what does the government

do why do you have a government
involved in in the research that we do.

Because it's based on a national vision
and a national motivation so the Kennedy

speech for going to the moon was a vision
motivational speech that really led

us that this country to become one of the
leaders of science right and to understand

the power of that the government has to
espouse a vision is more or less good

depending on the period you're in I want I
want to come in on comment about now but

I want to come out of a John Kennedy so
the 2nd thing that you do as a government

you allocate resources to your priorities
so funding for an I.A.S. funding for

research is one fight that you have to
find as members of the academia because

if you don't then the allocation of
resources by the political system

will be insufficient to sustain the dream
that you see on the left side so

that's the 2nd piece that
you have to understand but

understand also one thing is that when you
allocate resources to national priorities

like I was when I was at the end I
chose arguing our voice as scientists

is the least potent
least potent hear me out

we don't have as much
influence as we think we do

you know who has influence in making
the budget strong the patients

the ultimate customer the ultimate
voter the ultimate person who benefits.

From the public point of view.

From the investment and so that's what
I think we scientists always forget

is that we think that it's
enough to be rational and

explain a good thing to government
the government will buy this doctor and

do it my experience in Washington
was different we had 475

interest groups you know Alzheimer's
Association the cancer groups and on and

on they were a lot more powerful than
the National Academy of Medicine

the National Academy of Engineering that
all of the scientific bodies that were

important but they were not so sciences
is sufficient reason is necessary but

not sufficient so you have to
know how to build alliances and

the governance system will be peer
review or or the Congregational process

is the responsibility of government and
regulation have to yank all of these

institutions that's what governments
do not academia not industry and

the public interest is really what
drives the fundamental policies

of the government 1st defense in general
health education the social needs

programs and if you look at the debates
when you listen to the debate if you

can listen to a debate right now between
them it's more like a food fight but.

It's not a debate and but whatever you
listen you see that it rotates around that

some people want more defense spending and
defense research and

other people and more social programs and
you're see that and

then if you ask yourself well how
long can a government really last and

historically are they long lived you know
they're not as long lived as universities

so the only 2 governments that
have in this live for 200 years or

more unchanged are the U.K. and the U.S.
They're the only 2 every other

country has had a change of government
through revolutions or other forms of.

Upheaval so
governments don't last that long and

they have to evolve with that then you
can you have endless tree and you say OK

well what drives industry what are the key
drivers of the 10000 foot level.

The role of Fender Street is to assemble
the human capital that's created here

to the physical capital view for structure
the plants that allow whatever they do and

the financial capital and the financial
capital comes from multiple sources

pension funds the street you
know Wall Street investors and

so on but that financial capital is
the way by which governments and

societies allocate resources to
what they think is valuable and so

the financial capital that you see
here is allocated by a system that

requires you to fill market demands
you don't nobody's going to fun and

history if there's no demand for
the products that they produce but

you have to do it such that the return
on capital is sufficient to justify

the investment you understand so it's
a let's just not dream I mean if you think

industry I mean people have this dark side
of industry than industry is really a bad

thing it's not it's really part of
the ecosystem part of the economy of

the country that we need to to understand
in details but what do they want to do

is always under regulatory and
and competitive pressures so

I was talking this morning
to some of the faculty and

it was an interesting to see that
sometimes when you look at for

example the 7 earing manufacturing
project how much regulatory

plays a role in terms of
decisional processes within

the industry why because the chief of
manufacturing in the industry I can tell

you their number one fear is that the
F.D.A. comes in and stops them from from

doing something because they've changed
the process changing the process.

Of manufacturing industry is
the nightmare of the heaven of

manufacturing if you think there are you
know pro change and Pro technology and

pro innovation not the actually
want to things to keep things

stable so that they can I says the market
because of the regulatory forces and

obviously complete competition so
those are the major major issues so

in government you talk about public
interest here we talk about.

Transmission of knowledge and

Creation of human capital Vale to talk
about regulatory and competitive pressures

with what they have and they're driven
obviously to apply science and technology

they're not driven to basic research
we're driven of basic research here so

if you look at the lifetime of these
companies and these industrial and

entities it's much more 25 to 75 years
look look back at the companies that you

had around the country for
those of you have our age range so

how come many companies disappear and
no longer here and even companies like

General Electric which have been around
for a 100 is in great difficulty today and

a lot of people predict that it
might actually disappear and on and

on Westinghouse and a lot of big names
that you knew before have gone away

why because it's really hard to maintain
this regulatory competitive pressure with

a sufficient return on your investment and
if you don't get it you go away so

there isn't that sustainability
in industry as you do in and

it needs to be refilled and replenished
with googles and Amazons and and

then the next wave in the next wave of so

that's why so then when when you've heard
that in these sort of extracted what

the eco system looks like what I'd like to
do now is talk to you about the $1000.00

foot view about bio science which
all we are by scientists how

is this evolving in this ecosystem how is
it evolving what's driving bio science

today that you need to be conscious
of I think I can summarize it and

2 words complexity and precision.

So you say well what does it mean by that
how expires I'll tell you that there's

a tension between the enormous
complexity of biological systems you

know in 170 but
remembers there was a war on cancer and

the dream was that we would come up
with a magic bullet to treat cancer

people thought in those days that biology
was simple enough to be accessible

to a single target single single
bullet approach a magic bullet but

we know today that is not the case at
the same time because of this complexity

because of the stochastic
nature of cell replication and

the way things go well each one of
us is a very very unique individual.

I can guarantee you your biochemistry
is not the same as your neighbor your

fingerprints and not the same your iris is
not the same even your neurons by the way

you look at your neurons and you do
sang a single cell sequencing of neurons

you find actually a lot of
diversity between your D.N.A.

between your own neurons let alone between
200 different of individual cells just

like your fingerprints are unique unique
to you your brain circuitry is probably

unique to you as well statistically
it's almost having 2 human beings big

it being exactly the same is as unlikely
as winning the Mega 1000000 lottery

you know it's actually much
more likely than that so

that is the resolution of things we need
to characterize the individual but then

we need to characterize those individuals
in the context of a complex system so

those are the 2 major polls
that I have in my mind and

then when you understand that you
understand that there is no chance for

us to succeed unless we understand that we
need multiple disciplines coming together

to really crack these problems I mean
you need qualitative you need you need

a new method technology you need a culture
of cross disciplinary collaboration's and

you need structures to do that and
that's one of the things I did when I was

a Hopkins I created this that is without
walls because departments tend to put

barriers to introduce minority work and
I D N I which if you look back and

say you know it's been impact for the
number one thing I did was to create open

doors to the physical sciences biological
sciences and multi P.-I grants

which allow him to dismiss the work and
the convergence between

the physical the biological social and
data sciences is obvious it's

growing but it has profound implications
on your career so I'm talking to

the students here this is not empty talk
this is something that will impact you.

And you have to understand and

I'll tell you what the rules of
the road are at the end of the speech

the other thing is we have a huge
failure rate I mean I was shocked

you know when I took this job I thought I
was going to look like a hero you know.

Like I was so smart I was going to get
drugs done and treat Alzheimer's disease

in 2 years well wake up
the work of research and

development is the set with
failure you have to manage

failure not success success will come if
you're lucky failure will be there for

sure it's the normal state of affairs so

you have to really understand that the
knowledge is insufficient at this point

the knowledge is insufficient at this
point for us to be predictable and

predictive of the success of any
scientific approach which means that you

have to create the opportunity for people
to try and to to to do new things and

that's why I think basic research still
remains to me the priority of funding by

the government so if you look at medicine
and I'll take you there for a minute

if you look at medicine traditionally what
have we done as doctors for years and

years and years we waited for the patient
to be sick and then we looked at them and

you had big Gailey in
all of the artists total

all of the ancient doctors who
basically had these ideas and

they formed ideas about humor
fire heat that they were observed

from the outside it went from the outside
in and then you had organizers and

then Harvey discover the circulation in
the heart and all in all it was really

an outside in exploration today in
my medicine the world is inverter.

We are going from an inside out
exploration So what do I mean by that

we've become very cell centric and D.N.A.
centric and molecularly driven so

when you look at the way we look at the
world today think of concentric circles

and D.N.A. is at the center and in our
in the in the proteins in the cells and

then making it simple here which then
organize themselves in tissues and

organs and then they organize
themselves and to greatest organisms

which are interacting with behaviors
in our case because we do have emergent

behaviors due to all these changes which
are also driven by the environment so

all these layers of complexity and
the multiple

part potential permutations of
this complexity also drive this

complexity drive the individuality
that every one of us is unique but

to be at the D.N.A.
level when we did you know make studies or

the R N A level the protein level whatever
it is I mean at some point you will have

a divergence that occurs within your own
development to make you different than

anybody anybody else that makes sense
OK so if that makes sense then you say

OK fine how do I then play my cards
as an institution like endurance and

to focus on making you guys successful and
you being successful and

the way I look at it is think of it
as a multi-layered network system so

from the engineering of that or
guys here but basically it's a multilayer

interacting network so if you look at
the D.N.A. in the 1st place the D.N.A.

is interacting with its
transcription machinery it with its.

You know mix and then interacting with the
R. and a layer with the protein layer and

then making it simple in
the metallic layer and

in the cell of the environment so
that's what it is so

when you have a mutation in the D.N.A.
that mutation is affecting a network.

And if it's a mutation in
the in a molecule like C.

which represses a and and
encourages or repress is at the end

what you end up with is a network where
A and E. are going over x B. over Express.

Right the total thinking of this
approach is called Systems Biology so

people are trying and in many many
different fields yours included

to truly template that and so
is driven that obviously by a pigeon that

exceeds those mature lation isolation you
don't know the genetics of the genomics

the transcription of the translation
regulation of all of this and

the signaling of molecules across and
then obviously proteomics and

this is just one module you
know you can imagine this being

immunology maybe and then you have to
repeat the what at the metabolism level

the whole thing means you have the balance
that you need to have in the normal or

in the patient with diabetes and
all of that is essentially

leading to emergent properties that come
in from each one of these networks to

the next one to the next one and it's
really hard to invert that to understand

that in the very most even when you
study in our new neural system E.G.

These are a good way to get signal but
you can invert that into understanding how

the brain works so we have multiple
levels of doors that you can only

traverse one way so having said
that the environment is very key

we are not evolving in the vacuum where
it will be in the environment and so

you can never never forget
that at any one time

any one of these networks can be
affected by environmental factors.

All of that is the the those are the
elements of this complexity that I talk

about and you need to understand that
not everything is driven by genetics

because there's a sort of gene centric
view of the world here that everything is

in the mutation that you can try and
that's not true it cannot be true and

the environment drives a big
role actually to prove it

I'll show you an experiment we did we have
the statue of David that has no D.N.A.

I moved over to the U.S.
And this is what happened after one year.

I was.

So that was that was the proof positive
that the environment does play

a role as you well know we do have
an obesity crisis in the U.S.

But anyway so when you do analogies you
try to do analogies just a little bit like

if you were a extraterrestrial
scientist a 1000000 years from now and

you discover the chip on the on the right
side there and you try to ask yourself how

did that work you could probably discover
decomposed the chip in its components

you know transistors and all this but
you will never understand how it did.

You know a Power Point presentation or

a work perfect document you'd
never understand how the software

was unless you have a fully functioning
system so reductionism of taking things

apart is not sufficient you need to really
reintegrate and if you look at biological

sciences we are right at the knee of that
effort where we're trying to reintegrate.

The components into an understandable
whole and that's what cell networks

this is a cell in this on the left
side these are the back chemical and

signaling pathways that you can find and
the complexity is probably greater than

reading a chip from Intel
as we could imagine so

that's what I call the Iran that we are in
is the year of quantitative remember

the work quantitative there's no science
without you can't measure it you can't can

manage it and so and
it's a network type of biology So

having said that you say where are we
going we're going really from what I

call the hardware periods of science
where you take the components and

you try to understand each component and
how they work internally.

Ion channel a transcription factor and
whatever and

you're trying to put them back together
into a coherent system actually

Bruce Alberts uses the word the cell
as a computer if you listen to him and

I we've had this conversation many times
he believes the cells are fundamentally

computers and they're computing multiple
inputs to create an output across multiple

networks so for him and are sending the
software is as important or more critical

than him standing the hardware today so
that's what I think the molecular networks

the disease pathways the regulation in
health and disease that's the core of what

we do that was the core of the roadmap
we did the road map we said networks in

pathways scientific teams of the future
which have to be multi disciplinary and

really hard working together and
that should lead to a reclassification.

Diseases and if you can reclassify
diseases is a mechanism you can have

biomarkers in immunology or somewhere else
you can understand the drivers of health

and more importantly then you can
start to have a I play a role in

understanding the complexity across
multiple individuals because that's

the issue the issue is one sample of one
does not tell you what the truth is but

sample of the many without individuality
also doesn't tell tell you anything so

those are the factors and that's what I
said to people you know when you look at

disease networks and I'm just drawing
this one you can discover one but

some molecules are more important
than others that are molecules and

so the idea is to really do multi
targeting where you could the centrally

act on this molecule here which will
affect this one this 2 diseases and

this one and then you can combine it
maybe was this one or this one and

you have to decide which ones
are going to be the most important

you know what I'm describing here that's
what's happening today in immuno oncology

everybody non-college is
trying to figure out the next

partner to a P.D. one checkpoint inhibitor
that's the song that is exciting a lot

of people it's no longer the science of
the tool is the science in the system so

you can have an R.N.A. or a small molecule
or gene or cell therapy like you're doing

here or mono or multi specific antibodies
of peptides the idea is more the targeting

of coherent components of the network
that can affect the disease process.

So that's where we are and I'd like to
just give you a quick example of my own

field because I'm talking about things
that I don't quite know except that's

my field I know that and I can tell you
the impact of the imaging that this whole

revolution has led to is truly something
that I think I can share with you because

before we were imaging people now we
can imaging cells molecules but also

populations and I'll show you examples
of that but I think the key thing for

those of you are in imaging at all levels
is what's the definition of bio imaging So

here's my definition extracting
spatially and temporally resolve

structurally from core form and
functional biological information from

molecules to human right that's my 1st
work was actually to try to characterize

cancers from benign disease by
measuring count him content and

then I transfer them to osteoporosis
that was my 1st start up was to help.

People treat to process and
prevent surgery for benign disease so

these are the kind of things
that essentially drive imaging

it's different than a destructive
technique where you separate components

you don't know where they came from
here this is especially localized So

what are the 2 things that
are driving imaging sciences today

I'll share them with you one is
the revolution in molecular and

cellular imaging because we understand
better the biological targets too and

with antibodies with other molecules
we are able now to image things

that we didn't think we could
image even 5 years ago so

that example is when you develop
a particular molecule for

the human body you'd like to know where
it's going what Target is it attacking.

Isn't there in sufficient
concentration for

you to do something about it these
are the fundamental questions that

any scientist will ask themselves to then
you can image that you can image that.

With longer acting.

Longer acting isotopes
because you need to out and

localize and assess the functionality
of these molecules and

their targets in vivo you
need to characterize and

quantify the disease specific
patterns a given example and you have

checkpoint inhibitors right very expensive
but 50 percent of patients don't respond.

75 percent of patients don't respond why
and why not is it because the target is

not there and you would like to know if
the target is there right before you spend

a huge amount of money on this that's
a key question in immunology in ecology

Why is it that this is happening and
so multi modality imaging for

biomarkers has become is the revolution
that's ongoing So if you look at for

example this study that we did at Sanofi
with a partner in the Netherlands is it

corny I'm labeled monoclonal antibody that
we developed against T.G.F. beta now T.J.

of beta you know is a myth suppressor So
we wanted to suppress the suppressor and

but to do that you have to find out if to
give better is overexpressed in the target

because if it isn't you're going to
have side effects with no benefit and so

by doing this we showed that the the tumor

here was actually hot in terms of did
you have better and that's in humans

that's not an animal models it's directly
in humans and it really helps you guiding

the the therapy the same experiment
here different with different.

Tools this is the P.D.L.
one antibody against the P.D.L.

one target to find out of the cancers that
you see in the liver here in the lung

are actually active in P.L.O.
So you can use a P.D. one.

Moment to clone it again this is P.D.

and one in non-small cell lung
cancer with the Nano body so

this is a model clone all 150 kilo dolphin
very large it doesn't go everywhere

now nobody's This is from a company
in Belgium that developed

camel antibodies from camels and
llamas and they're very small

that they're very specific and they get
to the target very well and then you have

other molecules 14 killer Delton or
peptides Why is it important why is it.

Important it's absolutely critical
because you will not be able to

communicate all
the experiments you want to do

without having it in human biomarker
right otherwise we will treat

people with 56 drugs a huge cost
not knowing that they're efficient.

So here is a study by one of my students
money pumper of Hopkins who found

an agent that he could target against
pro trust a specific membrane antigen So

that's an integer and
that is specific to the prostate and

he was able to find a way
to to label it with 18 for

3 or in 18 and if you look at
the experiments does it in 3060 minutes

which is where within the half life of the
in 2 of the isotope and look on the left

side this is a patient who basically
perhaps had one lesion here but

these are with fractures and basically you
could have told them you do very well your

cancer is actually under control you have
a single lesion we're going to radiate

that you'll be doing fine but look what
happened when you did the specific imaging

with the P S M A this patient was riddled
with met a static disease that we didn't

really see in the previous test so
that's important Here's another aspect

which would be of interest to you because
you're doing Carty cell manufacturing So

Marty had this idea again
Marty's very creative guy he had

this idea of when you do a car to
sell you know you put an antibody

that targets the tumor right so
he said well let me put the P.S.A.

may let me put the into gin that targets
his marker on the same Carty cells so

he did a cartoon cell
experiment where he had

targeting a tumor in the targeting of
the imaging agent to understand and so

what he shows he said because the big
question in Carty cells are they there

are they going to the tumor or
are they doing what they're supposed to do

and so he showed that actually this is us
today 12 he was able to show that these

cells actually were infiltrating
the tumor in the mouse model so

that kind of revolution I think is going
to grow something you should look at

a lot of departments are not
looking at that in the M.E.

in the imaging departments because the you
need cross disciplinary collaboration is

with chemistry with radio chemistry
with nuclear medicine labs so you should

do it the 2nd revolution and I think
is important is what I call radio mix.

And this the machine of mental diagnostics
and I think it makes a lot of sense that

we're going to see a machine learning
matters classify these individual

patients occurrence you databases as
a radiologist when you look at the M.R.I.

of the brain you compare
it to your own memory but

your own memory might be 5000 cases max
10000 if you really really really good and

work every day you may have 15000
cases in your memory in your mind

most of which are probably unproven and
you don't know what the study is and

you don't really remember what the
previous study was in the new poll at this

it's messy it's messy except for local
disease we don't really know how to read

diffuse disease and so if you want
to track the evolution of a disease

you'd like to compare it
to some database right and

these databases are going to become
the the money of the realm like this is

the gold mine because if you are a system
you can put 5000000 patients M.R.I.'s of

the head in the database that you follow
over the years Look how important that

will be when I read the killer scanner
put the scan in the machine and

the scan will tell me this
patient is like Patient X. Y. Z..

In that database which have this disease
this disease that disease can you

imagine the power of that 10 years ago you
couldn't even have dreamed of doing this

we did it for mammography 20
years ago we couldn't do it for

the rest of the body as we do today and
so that information machine learning is

going to be fantastically critical
important which really basically you can

apply the exact same thinking to genomic
Dana the genetic data the single cell

sequencing data you name it any big data
scale can be approached now with this and

degraded reference data bases that
will teach us about populations

much more than they will teach us and
they will help us manage individuals so

this is for example the dream of all folks
in the imaging is to segment the image and

understand it so this is manual and this
is by did not work prediction it was done

by a fisherman with an Nvidia
platform amazing power it's

almost like you don't need to be
there outlines the the organs for

you and the nice thing is you
can use that to characterize.

Tumors and use can say well your liver
lesions is 123 they're different right and

here there's a very they go on
the machine actually can extract them and

analyze the texture of it to the point
where it can actually distribute through

these parameters that come out of the
periphery and separate what we call focal

NAJERA hyperplasia Hipparchus or
carcinoma a benign tumor or

a normal tissue so you can see how
the power of these 3 dimensionally.

Spread parameters which frankly I don't
know how they came up with that I don't

think anybody knows how the Al Gore and
the neural network did it they did that so

you can imagine now that if you want
to treat a cancer patient and you have

the sort of imaging that our routinely
achievable play this is a patient

with you can see the heart here is in
the liver you can see the stomach with

full fluid you can see the vessels going
into the liver and you still don't lesions

right this is a patient with a Cassano a
tumor which invited and they saw the liver

you know I would pay a fortune if
I'm really trying to understand and

treat this just to know very
quickly in a matter of days not

weeks not months if the therapy and
using is working so

how do I do this today basically to
grueling you measure one or 2 and you come

back and this is what we called the recess
criteria I spoke ologist will tell you.

When you this hear about this racist
criteria you can't resist laughing because

it's so but that's the only thing that
is working today that's what the F.D.A.

approves So you have to know that to
know the power potential power of this

now when I was at the end I actually
started we put an initiative in Bob

probably remembers that this is
an initiative between the industry to

understand Alzheimer's disease the start
of that in 2003 and we had M.R.I. images.

C.S.F. by sampling and
the need was to follow these patients and

Dr Thompson at U.C.L.A. had a really
brilliant idea where he would end became

the variation within the brain of patients
so green means not a lot of variation

blue a lot of variation in patients
between Alzheimer patients and

our patients right so he was able
to sort of use these indicators but

the next step is not that the next step is
going to be what I said knows we're going

to create huge databases of
normal development of the brain

this is a study that Judy Rappaport
started that he and I had in 2005 or

6 of trying to understand
normal brain development so

she would take images with M.R.I.
with the vision tensor imaging in all this

that sequence of times with these children
and follow them over years so it's like

a Framingham Study of the brain Framingham
was 30 the heart this is the study

of the brain and over time hopefully
will have reference databases for

normal development across
the ages all the way to 85 and

90 and see if we can identify
as early as possible

predictors of degenerative disease or

Parkinson's or
any other disease that's affecting us and

I think that's going to revolutionize that
because imagine if you could do this for

all human brain studies and put them in
the cloud and and have them and compare

that the power of that on biomedicine
is going to be a revolution so I think

I've made the point because it allows us
to be more precise and more predictive.

Imagine for example you do a colonoscopy
patients to prevent colon cancer every

535 years imagine if you could use one
of these imaging scanners an ob in 2

minutes you can get the entire body I mean
it's amazingly powerful and let's say you

had that as part of your physical every 5
years just like you do your blood test or

your cholesterol how powerful would that
be in terms of detecting disease in terms

of if it's combined with circulating
D.N.A. they could biopsies or anything

like that you can imagine a world where I
think all of us as individuals will have

the ability to be compared to normal and
abnormal to predict what we need to do or

not do to treat our patients so
it's not a pipe dream I don't know.

Sometimes people say Are you crazy it's
not never going to happen is too expensive

that's not true when I did the city
scanning at the beginning people thought I

was crazy because these machines were
costing 23000000 dollars at the time but

today you can get a scanner for
$300000.00 and

you can do more patients than ever to
the point where there are companies

that are offering this sort of test
right now so it's not a pipe dream

Now let me finish if I could by some
personal advice to each of you.

I'm talking to the students
here Bob not you.

You don't need any advice.

In this.

So the 1st thing that I would say is I
have some roles that like to share with

you 1st this 5050 rules
what does that mean.

As I was doing what I was doing I realized
that I was very curious about things

beyond my own specialty so
when I was reading radiology and

imaging maybe 50 percent of my reading was
really all Gene imaging but 50 percent

of that was actually non radiology non
imaging I would read about chemistry read

more like a biology or have science nature
things that I barely understood but over

time it made me richer in understanding
what I could get help from so

5050 new friends you know I had friends
in medicine and radiology at me and

me but I also made friends in other parts
of out of Hopkins one of my best friend

was actually an accountant it's
hard to believe it's true.

It's actually useful for taxes.

And the only way the only rule that I
would say $5050.00 is when you're in love

it's had a it's got to be 100 percent
this is both at the risk of your life and

all of Mr Musgrave mess that up so
that's 1st rule and you understand and

rich yourself in areas adjacent
to who you are the 2nd

is always the finer core
strategic team for yourself and

that's a hard one that's a hard one but
follow me a little bit and

what I've found but
this is retrospective and

I learned it after the fact I didn't
really know it at the time but

I think it's a great lesson that I
learned between all these fields.

If you really look at my
career scientifically and

you say what's the core theme what is it
the what is the strategy you followed what

is the theme that you were after I would
say that my big driver was to bring

qualitative sciences to the qualitative
biological sciences as I saw that.

So I could tell the radiologist was
looking at the the making judgement

with no measures enough
that's why create a C.T.

does a cometary a create a cardiac
tagging to measure cardiac

function we looked at M.R.I. signals for
tumors and but it was the same idea it

was how do you quantitate what really
happens in a biological system so

you can make sense of it then when I was
in and I was a little more mature so

I knew what I was going to do because when
I was asked to do the and I had the same

problem I had a Hopkins department we're
not talking to each other I mean don't get

in says he or he knows I mean we had
B M E but really reaching out to Don and

electrical engineering computer science
mechanical engineering was not natural so

we force ourselves to do it and
we created this this is without walls.

And so when I went to the N.H.S. I have to
say something that makes a lot of sense

to everyone right the core team and

the core to my idea that I came
up with was focused on what and

I should do and I as an institution
that no single institute could do

that was the exact same thing I did a 2nd
telling the department chief Well is there

anything that you would like to do yeah
I'd like to do this but I don't want to do

it I can't do it don't have the resources
and then when you surveyed the faculty or

every one of them would say well we need
a core of this or the core of that so

things that the institution is to do but
that no component of the institution

wants to do or can do so that was
the theme of the entire thing that I did

if you look back everything
came out of us this concept and

when you ask the question the people came
together the people that I said yeah and

then some not some people hated
the fact that you had the directors

are telling them what to do but I wasn't
telling them I said you decide and

once they really decided
they loved the component to

the idea that they could fund things
out of a common opportunity fun so

that's what we did you know
the structural biology the network and

Imus's the teams of the future of
multi-disciplinary teams now in my last

job I told people I said look I work to
last because you can't have a strategy if

you can write it in one sentence or 2 you
have to have your own personal strategy to

do that you have to define yourself so
I said that I don't

know if we were going to understand
molecular networks of disease and attack

them with multi targets approaches but
that's what we did for the past 7 years.

So the other thing that I can tell you
is that don't think the future is not

readable it's readable and you can tell
what's going to happen in the future

I'm surprising you're right everybody says
it's really hard to make predictions about

the future except in biomedicine in
biomedicine things take a long time so

what I found out when I evaluated and
programs I found out that when you looked

at the medicine of 20032004 everything
had been there the drug had been there

imaging had been there 30 years before
except that it was not appreciated so

the key thing for
your is what do I need to do to appreciate

what the power of something that
is currently here will be in 10 or

15 years and it's easier to do and
things are going to be in 5 years so

we know bio processing is going to
be very important in 51015 years.

Only a point so you decide what
is really critical for you.

And the lesson I always tell
people is that you know

surfing is a great sport on the surface
you probably don't stop around here.

But anyway if you did you know you know
that when you serve the the things that

make you makes you great is not your
skills as the waves that you pick so

pick the right wave that will drive you
to a future where you have a core theme

strategy that applies to the 5050
rolls Why is that because I will

leave you now the last concept
that I believe is important today

because everything I said as
an implication or how you develop yourself

as a student is completely different
than it was in the past so

in the past you know if you went to
a discipline you develop an expertise so

think of it as a vertical vertical
thing you you go down into your

silo deep deep deep deep deep in you're
really really good and you get a Ph D.

and you get a post doc and
you stay in the same lab and it's great.

But the problem is that science has
evolved from a reductionist era

where you needed to do the components of
systems to an integration or you need

to reintegrate the system but you can't do
that all by yourself so you could do and

say OK I'm going to go lateral and
collaborate with everybody out there so

you could say all of my my strategy is
going to connect I want to connect to

people connect and
be a connectivity guy was very shallow and

everything sort of jack of all trades and

master of none that's like salesman you
know salesmen are like that they know

everything they can tell you about the car
and you don't want to be that either so

what is it you want to be you want
what do you want to be this a T.V.

abounds where you know how
to connect laterally but

you have a grounding in what you do so
for me I'm grounded in imaging

I really believe in the science I know
that science inside out but also connected

like I said because the 5050 year olds you
know and I'm pretty good at other things I

can fake it most of the time and.

That makes you richer and
in the world of today building teams.

Through having this sort of design in your
market of being able to cross barriers and

to remove barriers or to understand each
other at different in different cultures

and different environment is critical to
success because the science the scale of

science today is different than
what it was 25 years ago so

the key thing is that all of these
are converging the complexity of

the precision the diagnostics and
therapeutics into precision medicine and

never forget to have fun but
the right way thank you very much

a

bit of.

He.

Right.

You know so that's a very good question
it's a fundamental question you know

because when you do.

These correlations is correlations and you
can't really have a causation unless you

do a validation experiment right so
the key thing is this is a learning system

that is essentially going to point you
towards possible logical causation

that you have to test in
a statistically valid way for

an intermediary period of time there
are some theoreticians that say that you

can actually in a large enough population
start to identify control groups so

you can do control but frankly from
the statistic coal point of view is not

a cure and you have biases in the
selection of the population and so on so

I think deep learning in in
applications like the one I showed

where you're really following
the same patient population over time

may have actually the power to
to identify the stronger and

stronger correlations because otherwise
the correlations will weaken over time you

know just like you do like we were
talking yesterday about genomics so

so might I don't have the answer to your
question I think is the key questions 96

percent of observational studies prove
to prove to be either incorrect or

size to distinctly invalid 96 percent so

anything you read in the eye always in
fun you know people who drink coffee at

midnight are likely to be dying sort of
Sooner or people do Techmeme milk at 12

life and live longer you know I mean
that is junk Minot's they can use.

You know.

Just.

My thought.

So this.

Week.

This week.

But I.

Didn't.

Really.

Know me.

Much good at it.

Good morning.

I'll.

Be good.

To get.

So step one is to reclassify what you
call a disease because you call a disease

a heterogeneous entity is probably not one
disease probably multiple like dimensional

time or I mean when you look at
Alzheimer's diagnosis many of them don't

have enough time or they have something
else so the real point I was making here

remember was a big effort that
I think needs to be made is

to be reclassified disease into
their molecular phenotype or

or or reclassify the heterogeneous
cancer you're talking about I mean you

know in breast cancer we know already
that there is that her 2 positive and

astringent possible triple negative and
so on till precise reclassification is

going to help solve this problem
in a great degree the 2nd is to

study very specifically the non-responders
beyond the classification Why

is it that a patient doesn't respond so
in the European Union I was pushing for

a program that they accepted actually
several $100000000.00 to look

at the non-response in auto immune
disease because it's amazing and

you you treat patients with a little mean
disease and 3040 percent of the respond or

stop responding and so

understanding that had to originate is
important because it's probably a source

of discovery of new targets and
new pathways last but

not least is the signal to
noise problem and in the in

the in the signal to noise problem but you
have you have insufficient sampling and

because you have insufficient sampling it
appears had a genius to use graining and

so you need bigger and bigger sample size
and so pretty competitive collaboration

between different entities different helps
the cells to accumulate enough in any one

disease is going to be very important
to control the signal to noise issue.

So from the patient point of view
you know the thing that I think

is very obvious is that as we moving more
earlier earlier in the disease state and

we are able to prolong survival for

many disease states that the patients
don't have to play a more active role so

I used to call that P 4 medicine you know
it's more predictive more personalized

more preemptive and more participatory So
the intelligent participation

requires a fair interpretation of the data
and this is what patient groups do and

they do very well I mean to star in a rare
diseases I would say the best example

is the Cystic Fibrosis Foundation there
are 30000 patients that were suffering

that were dying very young today the diet
43 there's no cure there are no treatment

with no cure at this point and you can
see that the they are participating and

actively going into their care the 2nd
point that you're making in terms of

lobbying is that you know patients
are everywhere in there and

they truly truly believe in
the power of research and

the good ones they don't really
you know you would think that

they think that you want they want you to
do treatment experiments you know give me

some stem cells give us some of that but
most of them if you look at the strategic

plans now the quality and I helped them
with that you know when I was in I WAS IT

guys what do you want when informed
patients are uninformed patients and

what is the best and you realize that
you have to participate and help them.

And participate in them if
you look at the walking sins

Disease Association Michael J. Fox and and
you look at their strategic plan they're

better than the many strategic plans I've
seen in academia or even in an age so

so I think it's a it's what I tell people
all our best friends are not the industry

not the politicians not the payers
one the insurance companies and

the patients and doctors who are losing
a lot of that you know there's a lot of

intermediary bureaucratic decisions
that are embedded now in the health care

system which have lowered the power
of physicians to decide and

if you're if you know about what we call
authorization guidelines go guys know

what authorization means pre-authorization
when a doctor prescribes a drug

there are there is a formulary that
the doctor has to abide by and

the insurance company can deny treatment
if you don't really do something else 1st

so for
example if you have high cholesterol and

there's a drug that really reduces
that this is the generic 1st 3 months

to do a test use a 2nd generic and
and go through that before you did

it take the expensive drugs or
the practicing medicine for us.

You see the same thing in children
my wife is a pediatrician and

she told me that she was fed up because
she would spend half an hour with

an unknown health care provider or
coordinator trying to argue

why because she does pediatric
endocrinologist a 3 foot kid needs

a growth hormone you know in
the near puberty and she was very So

we are losing that not gaining in that
you know powerful relationship that

you describe is less powerful which means
that we have to really engage with our.

Arborist on that which is patient's right
I mean we're not here to do in medicine

we're different here because
of engineering but but for

patients you have to engage and
we don't we don't.

And you would be surprised how little
I'm surprised people when I say that but

you get some mention was maybe
10 percent of the influence but

a good to 80 percent comes from
patients so you have to nurture them.

Yeah.

I didn't understand the last 2 parts and
the last 2 sentences.

I was a minus for the program managed.

It.

So it started 1st by you know I didn't
do the Road Map out of nowhere so

what I did is I was appointed in May and
the Irish had done had a director for

a couple years so it was really
frustrating to Congress because Congress

felt that they were increasing the funding
for an ice but they had no plan so

it's like me giving you money so I'll tell
you tomorrow I want to do with it and

they were frustrated with that they wanted
a plan they wanted an understanding so

what I did is it was different than
most directors do I didn't come with

preconceived idea I had some and some
don't mistake it I did have some ideas but

what I did was I invited 125
finalists from around the country and

I had joint sessions between the internal
as to director assisted scientist and

the external trying to identify just
what I said What is it that the and

I should do that no institute can do.

And we came up with about 97 different.

Different concepts and
ideas one of them that came actually

you'd be surprised we came from
the Nobel the group al Gilman who was

consulted Harold others who
said we have a problem.

Greencard was a big advocate of that
David Baltimore I mean they were

basically saying look you know it's
we're doing all these basic research and

you have all these clinical research over
here but nothing is happening in between

with the new emerging science we
need people who understand that and

can translate that into the clinical So
if you remember

that time we had the general clinical
research centers I happened to be a T.I.A.

of one at Hopkins before I went to an eye
so I knew how it works is basically

nursing support for the investigators and
you have a lab you have.

You know a connection to your lab and
pharmacy you know research pharmacy that's

what it was it was a service function so
what I think the folks were asking for

was not a service function but
a human capital formation of folks who

understood enough to traverse
the valley of death and

and really establish degree granting
programs in translational finds the 1st

one that was established actually was in
cancer at Memorial Sloan Kettering by

Harold on cancer science you know Panna
translational cancer science because you

do not see that I mean how who would you
go to if you have a do something here

who would you go to do a translational
effort you have physician scientists would

you know they're not in the best
of shape inside of academia and

you have so
they decide to go into basic research.

A lot 85 percent of them deficiencies and
up in labs doing doing

animal model research 50 percent and
up with the clinical part

of the late stage but there's nothing in
between and so that was the idea C.T.

essays do you train people in new science
do you really have them adopt and

transfer the basic research their
they participate in an understand

into a system that is completely different
than the previous one the hard thing was

to destroy the previous program because in
government you don't do that every program

was in many institutions 62 institutions
they have their own lobbyist and

the lobbyist I remember her I mean she
would look at me like I was the devil

you know what did I do to you
shall you around this and

that imagine I mean if you haven't had
that experience you should try it once

you know you get a chance to go to
government service you do it just just for

the education of it and so they didn't
want that until we found a way to to

get it done was proper convincing
of some Congress people and so

we switched that the key
thing to me is the is whether

they're successful is is what kind of
people are they training who then become

the translators between basic and clinical
which was the big gap that you found.

You.

Know if it.

Came.

Easy it.

Was.

The.

Then you know what the big problem
is not in industries in academia

if you if you really look at.

The ownership of data jealous ownership
of data comes from I could be a more than

industry if you if you read

I mean trying to get data from a group the
Framingham group I mean I don't want to.

But they consider that their
lifeblood if you do yellow just it is

this little lifeblood they
don't want to give the data so

the highest level resistance actually
is an idea not not an industry or

government government as you know we make
it public as soon as we have it right

the sequencing is a good example and
so on the way you see now

accumulation of data is in health
care systems so the partners for

example in Boston has a 3000000
database that we funded at the N.H.

that allows them to identify patients for
purposes the Geisinger clinic

there you have collaboration spree
pre-computed of an Alzheimer's disease for

example between industry and
those 2 arthritis so

it's coming together but frankly it's
it's haphazard I don't think there's

a systematic view I mean look up
which is an insurance company part of

United has a high 350000000
health records worldwide and they

do what they call real world effectiveness
which is the the understanding

of who is using what medication versus
what was this and comparing that so

they're not doing correlations that either
do comparative studies of I think a C.

and safety so some people
are accumulating that data already

it's going to be a huge issue the privacy
the protection of privacy and

it's just ongoing story but
I don't believe that this is going to be

a simple solution it's going to be
a case by case solution over time.

Healthcare systems are probably the most
likely to integrate data from their

patient base as they're getting larger So
there will be an Emory database for shore.

Thank you.