►
Description
The ability to combine quantitative and qualitative analytical methodologies to diverse patient information provides the opportunity to identify and classify patients suffering from a rare disease based on their medical history and response to treatment.
Panelists were: Bruce Cairns of UNC Chapel Hill; Michael Kowolenko of NC State University; Sandra Talbird of RTI Health Solutions; and Tara Britt of the NC Rare Disease Advisory Council.
For more information about the Data Science Roundtable series, visit bit.ly/SBDHroundtables.
A
Good
afternoon
and
thank
you
all
for
joining
us
today,
I'm
dr.
mia
shanley
for
executive
director
of
the
South
Seas
data
innovation
hub
for
those
of
you
who
may
not
be
till
you're
familiar
with
the
hub.
We
are
funded
through
the
NSF
National
Science
Foundation
and
our
host
institution
be
catalyzed
and
strengthen
public
private
partnerships
to
apply
data
science.
The
real
world
challenges
for
those
of
you
who
may
not
afraid
before
we
get
started,
I'd
like
to
mention
a
few
things
old
house
tapings
for
those
of
you
on
webex.
A
We
ask
that
you
you,
through
Mike,
unless
you've
been
called
on.
To
ask
the
question
of
tennis.
There
is
a
chat
box
in
WebEx,
which
you
can
raise
your
hand
and
tell
Carl
the
stuff
from
our
senior
project
manager.
Who
is
watching
the
webex
speed
that
you
had
a
question
to
ask,
and
then
we
will
fall
on
you
for
those
of
you
who
tweet
please
use
hashtags,
South,
big
data
hub
17s,
pdh
17
also
be
d
hugs
and
data
science,
and
also
you
can
follow
us
on
twitter
at
x,
health
data
hub.
A
So
welcome
to
the
big
data
hub
monthly
roundtable
theories.
We
provide
an
open
forum
for
exploring
emerging
research
questions
in
data
science.
We
identify
some
of
the
solutions.
Are
discussions
today,
we'll
focus
on
applying
artificial
intelligence
or
augmented
intelius,
it's
known
as
augmentation
the
diagnosis
of
rare
diseases.
A
We
afford
a
sea
wish
chalice
today
from
the
rare
disease
Observatory
project
they
are
recipients
of
the
National
Science
Foundation
Big
Data
health
planning
grants
we're
very
excited
to
have
them
here
today,
since
left
all
around
table,
as
featured
a
series
of
the
NSF's
big
data
hub
scope
and
sending
ground
projects.
With
that
I'd
like
to
our
panelist,
we
have
a
pair
of
Brits
who
the
associate
chair
of
the
North
Carolina
rare
disease,
Advisory,
Council
and
founder
of
a
non-profit,
NP
rare
disease.
A
Innovation
keep
co-wrote
the
legislation,
HP
823,
to
create
rare
disease,
Advisory
Council,
which
is
housed
at
UNC
School
of
Medicine.
The
direction
of
dr.
Bruce
Barnes
is
john
stackhouse
English
press,
professor
of
surgery,
so
the
joint
appointments
in
microbiology
immunology
at
the
University
earth,
Carolina
Chapel
Hill.
He
also
is
a
medical
director
of
the
North
carolina-based
Burn,
Center
elected
chairs,
back
to
be
inseparable,
and
then
we
have
dr.
Michael.
A
Cole
alenko
is
currently
the
director
of
the
Institute
of
the
next
generation
computing,
an
industry
fellow
in
the
center
of
innovation
management
studies,
research,
professor
in
the
Department
of
Computer
Science,
with
North
Carolina,
State
University
and
director
Vance,
next-generation
I
pieces
and
Sandra
Gilbert's
colored
Tolliver
Tolliver
is
a
senior
director
of
health
economics
or
tihs
she's,
a
member
of
the
newly
created
north
carolina
rare
disease
coalition,
based
on
her
involvement,
their
disease
at
the
associates
and
associate
trust.
A
nonprofit
organization,
you're
cute
asil,
a
diplomatic
Association
and
she's
also
here
to
represent
the
patient.
B
I'll
tell
you
a
little
bit
about
how
we
got
here.
Maybe
yes,
and
it's
something
a
little
bit
about
rare
diseases
in
general
I
keep
going,
though
the
statistic
rare
diseases
are
defined
about
one
in
ten
people
that
they
release
their
7000
their
diseases.
Collectively
they
expect
for
people
HIV
in
camp,
which
is
a
freak
staggering
statistic:
any
%
rare
diseases.
A
B
B
Fortunate
up
survive
with
sharing
king
and
I
wanted
to
write
cecilia
will
be
a
plow
I
got
involved.
So
I
work
with
researchers
that
you
insane
with
rare
disease
and
we
sort
of
meeting
with
patients
added
to
fund
research
pipeline
experience.
This
is
Taylor
before
she
was
diagnosed.
She
was
diagnosed
a
year.
Life
and
I
wasn't
bad
the
neurological
able-
and
this
is
her
before
and
as
we
were
educating
the
legislature
time
upon
this
rate,
adidas
either
account
said
you
know
1
million
of
your
constituents.
What
you
want
to
have
a
rate
of
these.
B
That
means
that
means
families
our
status
is,
we
need
affected,
so
I
just
wanted
to
be
mine.
People
the
day
day
is
great
and
it's
going
to
help.
Hopefully
the
project
report
he'll
help
Assad
you
go
diagnosis,
calm
and
accelerate
treatments.
We
talk
to
that's
where
to
sleep.
It
will
that's
why
the
data
is
so
important
for
this
project,
but
this
is
really
Pollard.
Excuse
the
quality
and
so
I
just
like
to
do
God
to
kill
people
a
little
bit
more
about
it.
B
C
Hello,
everybody
really
appreciate
the
opportunity
to
be
here
in
person
and
also
got
to
focus
on
line
so,
as
was
mentioned,
I'm
a
surgeon
I
take
your
firm
patients,
but
also
patients
with
severe
exfoliated
skin
conditions,
drug
reactions
and
so
forth.
I
deal
with
a
fair
amount
of
rare
diseases.
I
also
have
a
lab
and
they've
been
funded
through
the
NSF
and
student
medicine
and
nih,
and
our
approach
to
this
is
about
solving
a
real
world
problem.
That's
very
timely
and
very
challenging
talk
about
Big,
Data,
big
challenges,
but
also
big
impact.
C
What
do
I
mean?
Imagine
seven
thousand
diseases
that
some
of
them
are
known,
cystic
fibrosis
and
some
of
them
are
unknown,
where
we
might
be
able
to
understand
what
the
gene
is
or
maybe
not
asking
data
they're
mapping
the
human
genome.
This
whole
focus
on
precision,
medicine
and
personalized
medicine,
and
how
can
we
repurpose
drugs
to
figure
out
how
to
use
them?
Another
set
of
data,
the
drug
targeting
industry
in
the
FDA
and
the
way
you
develop
drugs
and
various
variants
on
how
to
get
through
the
FDA
process?
C
That's
another
big
set
of
data
electronic
medical
record.
The
patients
that
we
take
care
of
the
ones
that
are
diagnosed
the
ones
that
aren't
diagnosed
the
ones
that
give
pair
claims
to
go
through
Health
and
Human
Services
and
then,
what's
it
like
to
be
a
patient
in
a
family
who
has
all
of
this
economic
impact,
both
personally
as
well
as
in
your
family
and
in
your
community,
so
like
a
laser,
were
focused
on
a
relatively
straightforward
couple
of
words:
rare
diseases,
but
the
impacts
and
the
challenges
for
data
management
are
enormous.
C
With
valuable
about
this,
though,
even
though
it
sounds
like
a
chaotic
set
of
data,
is
that
each
one
of
the
questions
has
relevance
the
real
world
amp?
If
you
just
figure
out
one
gene
or
one
drug
target,
you
can
develop
a
medication
that
could
still
impact
thousands
of
people.
Maybe
you
could
develop
a
better
strategy
for
drug
approval
and
the
whole
idea
is
that
we-
we
not
just
look
at
this
as
a
way
of
analyzing
data
and
trying
to
bring
people
together,
but
we
transform
the
way
we
look
at
data.
C
We
need
people
who
think
about
data
systems
to
think
about
big
world
problems.
To
think
about
climate
change
and
other
things
that
are
so
difficult
model
and
what
kind
of
lessons
can
we
learn
to
be
able
to
address
these
problems
in
a
very
tangible
way,
in
a
very
specific
condition
for
a
very
specific
individual,
but
also
in
a
larger
population
level,
because
it
could
fundamentally
transform
the
way
we
develop
medications
and
the
way
we
treat
people,
because
how
else
are
we
going
to
do
that
with
limited
resources
in
the
future?
C
We've
got
to
become
more
efficient.
We
have
all
of
these
data
systems.
We
have
all
of
these
ways
of
looking
at
these
problems.
Arm
is
very
interested
in
it
now
because
of
the
opportunities
that
represent,
but
nobody
has
to
date
and
if
they
have
the
data,
they
don't
know
how
to
analyze.
So
that's
our
first
approach
and
and
just
my
three
more
minutes,
I,
don't
need
them.
Oh
well,.
B
B
B
C
Melanoma
rare
disease
answer,
you
have
melanoma,
it's
a
rare
disease
for
you
because
most
of
the
treatments
we
have
don't
worry,
so
it
becomes
more
than
just
about
rare
diseases.
It
becomes
about
commenting,
it
becomes
about
combinations
of
diseases
and
drug
therapies
that
you
use
it
becomes
about
evaluating
new
treatments
and
whether
or
not
they're
the
best
Nogales
and,
of
course,
as
we
map
families,
format,
communities
and
we
look
at
public
health.
All
of
these
things
become
integrated
in
how
we're
thinking
about
data
in
the
in
the
healthcare
space.
C
Well,
what's
nice
about
what
we're
proposing.
Is
that
then
there's
a
very
focused
agenda
on
rare
diseases,
of
the
interests
of
the
coalition's
and
of
the
NIH
and
a
pharma
driving
this
particular
element
of
this
question
forward,
so
that
if
somebody
can
create
a
political
enterprise
like
we
have
in
the
state
of
North
Carolina,
we
can
actually
have
the
wherewithal
to
bring
all
these
things
together
and
show
others
around
the
country
how
to
do
it
as
well.
D
I
might
go
away
go
away
during
this
once
ago,
in
looking
at
how
we
could
perhaps
apply
different
tools
are
dictating
analytics
to
this
problem
and
truthfully.
This
is
something
that
interested
me
for
many.
Many
years
before
I
came
back
in
india
about
eight
years
ago.
I
was
in
the
pharmaceutical
and
made
it
a
fairly
high
up
in
the
food
chain.
I've
been
involved
in
the
development
of
several
therapy.
D
It's
for
ear
diseases,
carpeted
back
to
rate
comment,
factor
time,
FC,
future
products
back
to
rate
alpha
1-antitrypsin,
those
sorts
that
were
really
helpful
with
James
lies.
It's
tremendously
important
that
we
understand
how
to
identify
a
target
and
how
to
identify
the
symptomology
of
these
different,
very
diseases,
as
it
enabled
us
design
the
appropriate
therapeutic.
Now.
One
of
the
issues
that
I
came
up
with
the
farm
up
was
the
fact
that
we
had
collected
tons
of
tons
of
data.
D
D
So,
anyway,
when
we
started
down
this
journey
of
looking
at
data,
one
thing
that
became
very
very
clear
is
and
I
sort
of
related
back
to
when
I
was
back
graduate
school
I
go
to
that
word
library,
Oscar
and
you
go
in
there
and
see
all
the
books
are
measured
if
I
know
how
to
ask
us
in
the
appropriate
manner
the
answers
in
and
I
started
to
look
at
data.
In
the
same
way,
it's
all
stacked
up
in
these
different
idols.
D
It's
up
to
us
to
figure
out
how
to
develop
a
process
that
allows
us
to
extract
the
important
information
that
can
enable
a
decision.
That's
really
what
we
want
to
do.
Data
analysis
for
the
fake
page
analysis
purpose.
You
have
to
enable
a
decision
process
with
that
is
who
has
this
disease,
but
can
we
do
better
for
this
person?
What
are
some
other
coral
that
we
can
bring
up?
What
are
the
targets?
These
are
very
pointed
questions.
When
we
asked
machines
to
ask
the
answer
questions.
They
don't
handle
ambiguity.
D
Very
well
right
ambiguities
with
a
lot
of
tests
with
the
tribe.
It's
with
trailer.
She
don't
do
that,
but
we
have
to
be
good
at
his
understanding
out
of
focus
the
question
in
understanding
the
facts
that
are
necessary
to
answer
that
question
and
machine
tool
ever
get
tired.
He'll,
never
stop
looking
as
long
as
you
in
the
right
direction.
Alas,
it
seems
like
they.
C
C
Relate
myself
of
Martin
I
here
so
from
a
political
perspective,
we
need
whatever
it
is,
that
might,
and
the
group
of
partners,
we
hope,
might
include
people
in
this
room.
We
need
it
to
be,
go
beyond
what's
known
in
the
books
and
be
able
to
tie
together
with
data,
because
one
of
the
biggest
problems
is
that
people
don't
know
what
the
recognized
I
don't
even
know.
C
C
Clear,
but
if
you
have
a
system
that
says
well,
you
have
this
team
problem
or
you
have
these
sorts
of
syndromes.
You
need
to
look
in
the
respiratory
tract
as
well,
so
that
there's
a
if
it
comes
into
it
is
that
these
these
processes
become
linked
because,
as
Sarah
mentioned,
if
you're
born
with
these
things
will
happen
and
be
manifested
many
years
later,
it
was
more
than
just
trying
to
manage
the
data.
We
know,
what
do
we
do
with
the
data?
B
D
It's
a
follow
up
on
what
Bruce
is
saying,
though,
when
you
sit
there,
you
think
about
how
we're
looking
at
biology
approach
basically
can
lead
is
just
as
biology
of
course,
and
mimic
it
in
some
way,
shape
or
form
and
how
we
do
our
analytic.
Now
we
look
at
connections.
No
I
can
start
out
by
looking
at
in
Oregon
I
to
look
Excel
like
a
look
at
biochemical
pathway.
I
can
look
at
genetic
activity
in
the
sense
of
transcription
translation.
D
Where
can
I
use
these
sorts
of
systems
to
start
to
make
those
connections?
Well,
if
I
know
pathways
and
I
know
commonalities
of
these
things,
then
I
get
started
to
my
interests.
Fortunately
collected
sort
and
reorganizing
information.
Thank
you
can
be
new
indications
of
what
they
happen
if
we
think
about
it
in
the
sense
of
triples.
Well,
you
know
again,
just
as
biology
is
really
good
that
we're
developing
systems
that
can
go
through
these
linkages.
D
So
we
see
that
we
can
really
do
a
lot
of
enablement
if
we
follow
it,
mimicked
what
we
see
the
body
doing
out
at
the
AIDS
and
how
it
responds
changes.
Remember
the
nice
part
about
biology
is
don't
like
students,
yourselves,
have
no
op,
they
do
everything
for
a
reason.
We
may
not
understand
it,
but
there's
a
reason.
So
if
we
understand
these
pathways
or
we
can
make
this
link-
which
is
data
and
understanding
of
these
fabrics,
perhaps
we
can
get
greater
insight
or
again
see
what
the
possibilities
are.
C
B
C
Say
fine,
I'm
good,
there's
an
urgency
to
get
this
done
because
the
people
who
are
affected
by
this
this
is
real.
They
don't
want
to
hear
something
is
impossible.
They
want
to
know
what
are
we
doing
after
we
working
on
it
and
are
you
going
to
help
us
solve
this
problem
and
so
we're
delighted
to
be
here?
We
want
you
to
hear
we
aren't
going
to
accept
it
camp
done
it's
impossible.
A
Care
is
an
extra
kid
that
was
also
a
good
segue
about
time
to
diagnose,
as
you
see,
I'm
standard,
Albert
and
researcher
RTI
health
solutions,
the
health
economics
division
there
to
work
a
lot
with
data
and
medical
research,
economic
models,
quantifying
burden
of
illness
of
various
diseases,
not
just
rare
all
sorts
of
diseases
and
building
xbox.
All
the.
A
A
A
She
was
doing
good
for
multiple
sclerosis.
Bution
a
family
member
got
diagnosed
pedestal
because
it
is
the
way
it's
inherited.
Autosomal
dominant
means
that
it
runs
in
families,
make
sure
you
have
a
six
percent
chance
of
having
catapillar.
If
you
have
it
passing
it
on
to
your
children
with
a
point
a
chance,
and
so
then
that
meant,
if
she
has
it,
that
this
person
may
have
it.
You
know
long
story
short
I
have
a
large.
You
know
40-person
family,
every
person
in
that
family
at
fifty
percent
chance.
B
A
B
A
People
in
later
in
their
life-
and
you
might
not
even
know
it
so
that's
why
I
got
involved
in
rare
disease.
Some
of
my
family
members,
co-founded,
a
patient
advocacy
organization
called
for
codicil
other
there's
one
other
patient
advocacy
organization
in
the
US
and
then
a
lot
of
times
really
is
a
path
for
people
who
learn
I.
Have
this
disorder?
There's
new
treatment
in
castles
case,
there's
very
little.
Research.
B
A
Know
what
can
I
do
I
need
to
make
a
different
times
have
to
do
that
and
a
colleague
of
mine,
Sharon
King
that
you're
introduced.
Also,
she
was
the
founder
of
her
daughter's
organization
for
baton,
speed,
Taylor's
tail,
so
this
really
the
patient
side
of
the
story
and
but
is
a
common
path
from
people
later.
A
B
A
B
A
And
so
when
Tara
said
this
is
the
question
I
want
to
answer.
You
know
I
want
to
answer.
What's
the
impact
of
rare
disease
in
both
Carolina,
how
many
people
are
affected?
Okay,
North
Carolina's,
10
million
people,
so
we
know
it's
probably
around
a
million
both
quantify
the
direct
medical
costs
of
that.
What.
B
A
C
A
B
B
A
C
B
B
C
A
So,
basically,
I
joined
saying
how
can
I
help
you?
How
can
I
get
data
I
know
it
won't
be
as
easy
as
Kara
was
hoping
it
would
be,
and
so
that
kind
of
leads
well
into
the
data
side
of
integrating
systems.
Nsf
planning
grant
is
for
to
see
how
we
could
integrate
systems.
I
was
trying
to
provide
data
on
the
go
to
the
next
slide
patient
registry
side,
which
is
so.
This
is
just
an
example
of
how
difficulty,
and
so
you
know,
how
would
this
project
help
patients?
A
Basically,
this
on
a
part
of
the
slide
will
improve
diagnosis.
It
can
definitely
reduce
diagnosis
time,
but
also
improve
research.
I
mean
patients
are
out
there
who
want
to
be
diagnosed
if
they're
on
their
guidance
is
Bernie,
which
is
care,
said
I,
think
it
has
an
outer
time
at
seven
years,
foundations
of
Defense
over
your
some
patients.
We
were
actually
just
talking
to
Dylan
diagnosed,
staying
up
there
dealing
with
their
disorder
and
they
don't
know
what
it
is
and
yep
where.
B
C
B
A
B
Whatever
they
were
working
with
conspiracy
in
Michael,
editing
like
reduced
sentence,
all
good,
it's
all
be
all
be
in
one
database
and
then
maybe
a
systematic
code,
especially
the
doctors.
That
really
don't
know
that
much
about
rare
diseases,
and
I
will
tell
you
from
her
work
with
many
many
many
patient
advocates
with
us
patients.
They
live.
B
Right
now
they
have
two
children
ages.
Five
and
two
rare
diseases
not
been
diagnosed
and
live
your
life.
If
you
amusing
and
the
frustration
of
every
single
person
I
meet
in
reinventing
the
wheel,
they
don't
know
where
to
go
they're
living
on
facebook,
who
is
not
really
a
simple
resources
for
make
that
part
of
what
is
data
so.
D
What
we're
going
to
take
the
approach
now,
the
soup?
We
really
do
have
a
wonderful
supply
of
people's
lives,
preachers
fire
here,
but
we
have
to
understand
those
depending
upon
the
type
of
question.
We're
asking
will
determine
the
type
of
tool
and
analytical
by
so
the
way
we
approach
the
problem
is
not
to
say
well,
we'll
do
database.
D
We
rather
go
back
to
the
point
I
made
earlier,
which
you
say:
look
if
we
can
frame
these
questions
appropriately,
we'll
figure
out
the
appropriate
tools
to
put
data
in
the
appropriate
structure
to
do
the
analytic
or
to
do
the
retrieval
now
this.
What
this
does
is.
It
allows
us
to
know
different
types
of
systems
so
that
we
can
tailor
it
to
what
does
a
physician
need
in
order
to
enable
an
understanding
or
better
insights
to
their
face?
What
does
Farman
need
with
regard
to?
D
What
are
the
mechanisms
are
pathways
x4
and
what
dissipation
is
creepy?
It's
regard
to,
who
are
the
supports
for
the
support
systems?
Where
can
they
go?
All
of
these
are
different
types
of
questions
that
require
different
types
of
input.
Now
what
we've
been
able
to
do
is
to
really
be
agnostic
about
where
information
comes,
provided
the
types
of
been
destroyed
bill.
D
So,
if
you
can
think
about
it
from
a
data
point
of
view,
we're
ingesting
anything
from
structured
data
in
databases
to
unstructured
text
and
figuring
out
ways
of
doing
a
series
of
buildings
on
it
and
what
we
start
with
is
generally
some
high-level
filters
that
allow
us
to
isolate
different
animation
and
from
there
we
can
become
more
definitive
and
how
we
do
the
extraction
in
our
specificity.
Contacts
now
and
then
involve
the
analytic
around
that
no,
but
we
probably
trying
to
do
or
over.
We
are
in
the
process.
D
We
not
advocated
movie
more,
very
well
as
with
an
exempt
park
and
that's
with
ccd,
which
is
a
primary
ciliary
dyskinesia.
These
active
affiliate
themselves,
I
very
interesting
item.
Billiards
is
the
world's
expert
is
at
I'm
UNC,
like
Knowles,
and
what
we've
tried
to
do.
What
we're
doing
is
we're
working
with
white
and
his
team
to
understand.
When
you
look
at
a
patient,
how
do
you
perceive
that
nation?
What
are
the
processes
you're
going
through?
How
are
you
doing
your
sort
with
regard
to
what
are
the
signs
and
symptoms
were
the
clinical
presentations?
D
What
is
their
hierarchy
of
decision-making
process
you're
doing
and
then
can
we
mimic
that
in
some
form
of
analytic,
either
first
through
understanding
and
sorting
the
information
appropriately
and
then
doing
the
extraction?
So
once
we
do
the
extraction,
then
we
can
start
thinking
about.
How
do
we
want
what
we
act?
What
we
have
at
the
end
of
the
day
think
about
it
is.
B
D
Have
a
matrix
thanks,
guys
limit,
then
I
can
start
playing
now.
I
can
start
doing
mine
that
one
can
start
looking
at
relationships.
So
the
idea
here
is:
how
can
I
take
the
transform
with
dr.
Knowles
feet
and
get
it
to
the
machine?
Is
that
all
that
stuff
up
for
so
it's
their
credit
in
a
nice
people?
This
is
everything
is
here
now
what
I
then
go
back
to
Daniel
actors
say:
look
if
I
see
all
these
things
lined
up.
D
What's
the
probability
that
is
this
disease,
so
in
essence,
what
we're
doing
is
filling
on
a
matrix
understanding
what
has
higher
priorities
and
other
addictive
that
matrix
and
looking
at
your
associations
the
thing
about
the
machine,
what
really
enables
instead,
she
is,
if
you
think,
about
a
patient's
at
30
years
of
medical
records
you
trying
to
find
some
medicals
okay.
So
that's
where
the
machines
come
in
very
handy
right,
not
going
to
tire
if
I
can
translate
the
way
at
the
position
thinks
about
the
diagnosis
code
is
the
triaging
then
I
can
make.
D
You
should
mimic
that
step
in
fill
in
the
blanks
for
where
the
machine
can
fill
in
that
can
I
apply
probabilistic
methods
to
see
whether
or
not
I
have
a
higher
or
a
lower
likely
of
a
disease.
The
answer
is
yes
and
that's
what
we're
building
now
now
the
nice
thing
about
that
sort
of
system
is
because
it
is
systems
ace.
We
can
move
from
one
disease.
C
So
I'm
very
practical
level.
You
know
this
is
a
planning
grant
and
we
are
appreciated.
We
are
enormously
appreciative
of
the
support
of
the
data
hub
and
we
really
are
trying
to
create
this
rare
disease
with
servatory,
and
then
the
idea
would
be
that
we
would
have
a
full
BD
spoke
proposal
that
would
allow
us
to
really
answer
these
questions
is
a
very
ambitious
agenda.
I
think
the
other
very
specific
set
of
questions
that
we
have
is
going
to
be.
How
do
you
even
approach
this?
How
do
you
prioritize
what
information
you
need
what's
available?
C
What's
not
available
and
what
will
have
the
largest
impact?
But
how
do
you
have
you
sort
of
triage
a
system
like
this?
The
thing
could
be
scaled
up
and
then
be
applied
across
the
North
Carolina,
but
then
the
other
South
BD
spoke
hub,
States
and
then,
of
course,
nationally
with
the
idea
being
that
we
would
like
to
develop
a
framework
in
the
beginning.
That
could
then
have
a
very
specific
set
of
goals
for
a
larger
proposal.
B
B
D
C
B
D
A
B
D
Lot
of
the
information
that
we
structured,
if
I'm
thinking
about
laboratory
I,
try
if
I,
can
add
subtract
x
5
in
that
we
do
a
lot
of
work
in
structured
text
and
where
they
kind
of
fortunate
enough,
if
it
involved
with
the
development
of
Watson
process
using
unstructured
text
panel
de
Lima
framework
and
worked
with
IBM
in
developing
that
our
community,
so
I'd
implemented
that
same
sort
of
strategy.
So
with
that
allows
us
to
do
now
is
go
into
medical
records.
You
pull
backs
out
in
a
pretty
straightforward
I.
D
What
we
do
is
we
do
a
combination
of
a
rules-based
system
along
with
a
little
bit
of
G
learning,
because
what
we
try
to
do
it
and
they're
all
different
philosophies,
but
my
take
is
that
I'm
going
to
introduce
a
little
bit
of
bias
beginning.
I
want
to
focus
on
known
data
sets
I
want
to
ingest
all
that
material
and
then,
when
I
have
those
results,
then
I'll
apply.
Different
types
of
backs.
Word
stretch
see
if
I've
missed
anything
to
bring
in
the
appropriate
page.
D
D
Those
are
the
sorts
of
tests
that
week
and
then
we
go
through
a
whole
process
of
understanding
how
to
build
a
annotators
by
the
annotated.
Those
sorts
of
things
now
I
take
that
together.
In
that
way,
conversion
of
the
annotations
into
a
numeric
framework
that
allows
me
to
line
it
up
in
my
structured
data
sets.
What
does
I
have
that
in
place?
Then
we
didn't
put
in
we
overlay
yet
another
matrices,
our
table
that
deals
with
the
physician
and
how
they
rate
or
score
different
activities.
D
So
now
what
I
can
do
is
I
can
run
things
like
classes
against
it
or
is
some
sort
of
multi-criteria
decision
making
algorithm.
It
allows
me
to
look
at
all
these
combinations
that
and
come
up
with
its
course
that
interns
is
okay,
the
output,
little
probably
that
the
information
that's
relevant.
So
the
types
of
data
we
get
right
into
their
medical
records,
so
we
also
go
in
and
do
like
PubMed
to
be
able
to
correlate.
D
You
know
this
is
one
of
the
way
from
testing
system
can
I
get
an
annotator
that
can
expect
all
these
values
out
of
a
thousand
papers
from
met.
That
would
allow
me
to
answer
questions
about
this
disease.
The
answer
is
yeah.
We
get
those
systems
enjoy
our
Trek
nation.
We
then
roll
that,
over
into
the
medical
record,
to
see
how
it
okay,
the
framework
that
we
use
it
I
said
it
as
you
inna,
and
there
are.
B
A
D
D
B
D
From
ana
performance
point
of
view,
when
you
start
dealing
with
very
large
indexes
and
trying
to
maintain
the
temporary
file
system,
sanaya
rates,
it
would
output
rates
what
we
found
was
it
became
impossible
to
have.
It
was
very
difficult
to
run
stateless
had
to
be
state
lettuce.
We
had
to
know
where
the
stuff
was
physical
locations,
being
tactical
back
and
forth
appropriately,
where
the
machines
would
die.
D
The
other
problem
that
we
had
was
the
number
of
threads,
and
such
we've
often
had
problems
with
each
size
of
Java
with
regard
to
how
much
memory,
so
we
developed
it,
and
this
we
work
out
with
IBM
truthfully.
This
took
about
six
years
of
work,
to
figure
out
in
understanding
how
to
get
appropriate
architecture
to
do
this
type
of
analytic
and
when
we
ended
up
developing
what
is
on
their
power
architecture,
then
using
that
live
with
their
ESS.
D
A
D
Everybody
wants
to
share
everything,
I
personally,
don't
understand
it,
but
I'm
out
there
we
can
get
it.
So,
no,
it's
a
matter
of
fact
in
things
like
where
diseases
will
go
to
the
flocks
of
the
patients
who
are
there,
so
we
get
a
good
understanding,
another
test
imitation
site,
so
they
need.
We
need
to
understand
what
are
the
white
space?
Is
there
with
regard
to
what
are
their
wants
and
needs,
and
are
we
meeting
them
appropriate?
D
But
by
doing
these
sorts
of
extractions
either
looking
logs
or
understanding
the
support
groups,
we
can
start
to
develop
a
profile
around
that
much
like
Amazon
does
with
you.
Do
the
same
thing
with
regard
to
what
are
their
concerns
can
be
directly
the
right
places,
and
hopefully
we
can
do
in
a
seamless
fashion.
To
me,
one
of
the
last
things
that
a
patient
requires
to
do
is
to
have
to
struggle
through
a
website
to
try
to
find
it.
C
If
I
can
just
add,
you
know
we're
so
excited
about
this
strategy.
For
those
of
us
who
have
to
go
through
medical
health
record
hole
I'm
now,
where
they
don't
talk
to
each
other,
Cerner
and
epic
demons,
and
and
and
that
I've
met
and
and
and
just
that,
the
wealth
of
information
that's
possible
much
less
what
the
patients
and
their
families
are
trying
to
feed
us.
C
There's
no
structured
mechanism
for
us
to
address
this
issue,
not
just
for
one
disease,
certainly
not
for
rare
disease
it
and
so
that
the
idea
would
be,
and
we
leverage
some
of
these
really
advanced
developments
in
data
management
and
input
and
really
create
a
framework
where
not
only
would
it
address
our
specific
issue
but
might
help
us
solve
some
of
the
siloed
data
storage
management,
interaction,
problems
on
a
larger
scale.
Well,.
B
B
Before
I
hand
insistently,
you
were
talking
about
you,
you
talk
to
an
unstructured
text
via
and
also,
and
you
talk
about
social
structure.
Commission
computing
mathematical
resides
you
include
in
in
there
beautiful
data
data,
that's
encoded
with
some
semantics
for
mythology
and
could
keep
more
directly
with
to.
D
Tell
you
the
truth
and
I
generally
get
trouble.
We
found
it
easier
to
build
our
own
semantic
system
that
focuses
upon,
because
what
we
want
to
do
is
have
a
very
portable
process.
That's
based
on
different
systems,
interactions,
we've
developed
our
own
and
we
found
that
act
has
become
the
most
flexible
platform
for
us.
Did
you
decide
now
when,
when
you
think
about
it
semantics
or
known
types
of
ontologies,
you
can
be
somewhat
restricted
how
its
information,
so
we
sort
of
you
did
as
well.
D
You
know
what
we
want
to
do
retain
certain
parts
that
we
leverage
and
bring
it
into
the
system.
Things
like
if
you
look
at
med
room
like
this
android
is
the
address
organ
system.
They
have
a
common
glossary
of
terms
and
how
they
describe
different
systems,
but
we
incorporate
that
it.
Just
we
incorporate
I,
see
agencies,
so
all
of
those
pieces
are
there
and
being
tagged.
It
brought
along
and
celtics
that
the
truth
table
or
matrix,
I
clock
so
yeah.
We
do
leverage
it,
but
we're
not
totally.
Okay,.
A
B
A
Definition
varies
worldwide.
I
think
I
ôm
has
adopted
a
definition:
wh
World
Health
Organization
and
adopted
a
definition
so
in
the
US
it
actually
has
a
of
a
strange
definition.
Most
of
it's
usually
a
one
per
hundred
thousand
people,
but
in
the
US
it's
actually
any
disease
or
disorder
that
affects
less
less.
B
Us
permission
because
you
all
were
sort
of
arguing
the
practices
aren't
arm
through
the
prayer
right
so
in
in
the
system
that
you're
working
on
or
envisioning.
Is
it
here's
a
disease
where
the
bunch
of
symptoms
we're
going
to
do
a
matching,
or
is
it
we're
going
to
turn
the
system
loose
on
a
bunch
of
data
and
see
if
it
comes
up
or
the
same
clusters
are
of
people
with
life
because
maybe,
as
I
think
some
of
you
routing
their
connections
that
you
know
humans
like
to
see
patterns
right?
Well,
as
you
see
the
eyes.
D
D
I
will
show
thee,
but
we
try
to
do.
Is
we
start
with
biases?
That
is
we're
looking
for
some
30
pins
and
if
we
don't
in
hopefully
we'll
see
that
again
will
shift
the
data
in
one
way,
shape
or
form
with
that,
if
the
nice
part
about
it
is
we
could
once
we
have
that
we
can
then
go
back
and
do
that.
Well,
let's
let
she
nav
at
it
approach
and
it's
that
no
because
it's
one
of
the
problems
that
I
basically
remember
I'm
an
immunologist
Anakin,
just
an
interesting
transition.
D
The
one
of
the
things
that
I
try
to
get
to
understand
is
that
look.
We
want
to
have
some
sort
of
context
and
there's
nothing
to
stop
me
from
having
this
joint
approach
to
let
let's
synergizes
one
thing
is
I
exclusive
of
the
other,
it's
sort
of
like
when
the
quad
people
say
they
don't
want
to
lose
quality
you're
not
going
to
get
a
good
responses.
B
C
Means
I'll
respond
in
even
more
ignorant
wages.
You
know,
as
a
physician
I,
think
that
that's
the
paradox
or
the
the
dynamic
that
you
asked
is
why
it
is
such
a
beautiful
question
from
my
perspective,
because
as
Mike
just
says
both
so
yes,
it's
rare
if
it's
a
rare
condition
and
it's
absolutely
common,
if
you
lump
them
all
together
at
7000
diseases
with
all
these
various
amino
type.
C
The
beauty
is,
is
that
you
can
really
try
to
get
in-depth
for
one
particular
conditions:
a
cystic
fibrosis
where
we
have
a
fair
amount
of
data,
but
there's
still
a
lot
of
work
to
try
to
figure
out
how
to
connect
all
of
this
and
be
able
to
come
up
with
some
useful
guidance
on
anything,
including
people
who
are
seeing
a
patient
in
may
or
may
not
have
cystic
fibrosis.
And
then
you
have
all
these
other
syndromes,
where
it's
really
not
entirely
clear.
You
have
more
and
more
data
coming
out.
C
You
know
now
that
we've
mapped
human
genome,
we're
understanding
more
and
more
with
these
mutations
are
for
understanding
other
aspects
of
the
other
way,
the
precision
medicine
might
work
or
drug
targeting.
All
of
these
new
developments
there's
not
a
really
effective
way
to
integrate
that
and
there's
not
a
structure
for
the
in
to
integrate
it.
So
that's
why
we
think
it's
so
exciting,
because
there's
any
number
of
questions
that
might
seem
on
the
broad
spectrum
and
diametrically
opposed,
but
for
rare
diseases
they're
all
relevant
to
each
other
and.
B
Ever
one
person,
or
talking
to
NC,
State,
Michael
and
John
about
this,
what
we
envisioned
is
it
will
be
in
malad
interventionism
like
a
portal,
there's
going
to
be
a
lot
of
data
without
it,
maybe
part
of
its
going
to
be
patient
advocates.
Can
access
from
critical
conditions
can
access,
researchers,
access
and
eventually
Armas.
A
B
A
B
In
fact,
I
am
kind
of
have
a
biology
question
and
it's
one
that
was
triggered
by
a
crisp
austin.
You
have
a
meeting
where
he
was
explaining
to
the
project
group
that
nobody
has
a
catalog
of
all
the
diseases
which
I
found
very
interesting,
and
it's
really
quantity.
So
I
guess
I'm
asking
you
all
who
have
much
better
biological
knowledge
than
I
do.
Do
you
believe
that
there
are
a
discrete
number
of
diseases
and
they
just
there
for
a
discrete
number
of
phenotypes
that
you
can
actually
biologically
differentiate
with
clear
so.
C
Again,
I'm
not
a
data
person
but
I
see
that
as
a
data
question
as
much
as
a
biological
question,
because
everything
is
a
continuum
just
like
life
age,
you
know
for
burn
patient
20
years
old
as
middle-aged
30
years
old
is
old
and
anything
older
than
that
your
survival
risk
is
very
low.
So
it
depends
on
the
circumstance
whether,
where
you
fit
into
a
continuum
but
then
at
the
same
time,
there's
discrete
patterns
of
disease,
phenotypes
and
potential
drug
therapeutics.
C
Because
remember
when
you
take
up
when
you
get
on
a
plane,
the
expectation
is
the
plane
will
take
off
and
the
plane
will
land.
That
is
not
true.
When
you're
born
when
you're
born
at
some
point,
the
biological
systems
failed
and
we
all
die.
So
the
expectation
we're
always
going
to
be
at
the
same
place
is
going
to
be
very
different.
C
What
we're
trying
to
do
is
figure
out
a
very
practical
way
to
sort
of
take
this
continuum
which
could
be
enormously
complex,
so
therefore
impossible
and
say:
well,
we
have
to
decide
because
the
vast
majority
of
you
I'm
I'll
presume,
would
rather
not
come
see
a
doctor.
None
of
you
want
to
come,
see
the
burn
doctor,
and
so
therefore,
this
is
not
relevant
to
you
until
you
get
sick
or
until
your
child
get
sick,
and
so
the
question
is
for
the
people
who
have
to
access
the
system.
C
How
can
we
use
this
new
burst
of
information
to
catalog
it
to
effectively
deliver
that
care
better?
So
I
think
it's
it's
it's
a
continuum
for
sure,
but
the
real
effectiveness
is.
How
can
you
take
all
of
this
data?
That's
just
really
confusing
and
and
give
it
hierarchical
nature
so
that
you
can
use
it
appropriately
to
make
a
difference
in
people's
lives
rather
than
just
have
it
be
boys
and.
A
I
just
want
to
clarify:
there
are
just
in
case
information.
There
are
multiple
large
databases
that
have
a
very
hot,
very
comprehensive
attempted
logging,
rare
diseases.
So
there's
orphanet
in
in
europe
there's
a
genetic
and
rare
disease
program
which
is
at
nih
national
institute
of
health
as
a
large
private.
A
B
C
B
C
The
major
players
that
are
really
affecting
your
ability
to
have
a
healthy
life
or
the
ones
that
you
know
you
can
intervene
and
make
a
difference.
You
diagnose
them
earlier
and
a
certain
kind
of
treatment.
You
won't
have
this
long-term
effect,
so
it
ends
up
being
I,
know
I'm,
not
terminologies
by
the
hierarchical
question,
so
that
what
answer
comes
out
is
not
just
information.
That's
useful
and.
D
That's
where
we
want
to
link
it
back
to
what
is
the
biology
is
that
to
me
is
the
way
that
allows
us
to
now
try
juice
to
answer
the
predictive
max
in
the
sense
of
if
I
can
find
if
I'm,
a
researcher
and
I
can
find
commonalities
and
a
crazy
like
that,
the
new
associations,
it
opens
up
new
possibilities,
and
this
is
really
where
we
want
to
take
cheap
trick.
This
is
what
they're
good
at
the
trick
is
in
organizing
the
information
and
making.
B
C
B
C
And
then
the
inability
to
figure
out
what
the
side
effects
are,
and
so
what's
beautiful
in
my
view
about
this
rare
disease
question-
is
that
the
enormous
complexity
of
it
makes
it
seem
like
there's
no
real
practical
answer
to?
At
the
same
time,
there
is
a
very
local
urgent
driving
need,
not
let
that
answer,
and
so
we're
at
that
interface
means
something
nobody
has
ever
tried
to
do
before,
which
I
just
wanted
to
say.
C
I
hadn't
had
a
chance
the
how
much
we
appreciate
being
a
part
of
the
ed
hub,
because
the
reality
is
is
that
if
this
program
didn't
exist
in
SF
had
created
it,
and
you
all
hadn't
been
interested
in
sort
of
these
kinds
of
large
good
questions.
This
would
not
happen.
It
would
not
exist
the
collaborations
with
NC
State.
You
know
talking
with
Michael
about
thinking
about
these
issues
in
front
of
you
and
thinking
about.
How
can
we
make
this
happen?
C
B
Able
so
can
I
ask
one
more
question:
I'm
sorry,
different
system
I
find
this
absolutely
fascinating.
So
you
said
why
rare
was
rare,
but
yet
comment
I
understand
what
you're
saying
but
I.
Don't
think
that
the
other
really
valuable
and
I'm
making
a
presumption
here
that
this
is
true.
So
please
correct
me
if
I'm
wrong.
The
other
really
valuable
aspect
of
looking
at
rare
diseases
is
that
this
is
like
I
Pappas's.
They
all
are
almost
certainly
linked
to
genomic
the
less
we
all
get.
C
C
Well
one
day,
one
and
one
could
say:
that's
probably
true
of
every,
so
you
know
how,
because
remember
that
all
of
these
biological
problems
are
related
to
development.
Development
is
all
related
to
genetic
information.
Genetic
information
is
shared,
amongst
you
know
almost
all
of
our
species,
it's
all
in
the
processing,
and
then
you
could
say
it's
all
and
then
you
know
what
happens
you
now
of
course
trauma.
We
always
say
you
know,
I'm
hundred
myself
trauma
surgeon,
there's
no
cure
for
stupid
and
we
don't
really
have
the
database.
B
D
The
issue
becomes
one
of
what
people
tend
to
do
and,
as
I
observed
to
my
old
life
on
my
boyhood
biologists,
they
would
look
at
their
one
cheek.
All
right,
very
deeply
didn't
understand
where
they
were
looking
at
was
one
snapshot
inside
of
what
was
going
on
with
the
genetic
control
systems.
Okay,
one
gene,
one
snapshot:
we.
B
B
A
B
B
D
There,
there
dear
very
similar,
basically
what
they've
done
if
they
integrated
machine
learning
aspect
of
things
head
to
their
system.
Mine
I
have
to
annually
from
one
system
to
another
system,
but
at
the
end
of
the
day,
we're
doing
the
same
thing
I
mind
what
we
did.
They
are
a
bit
more
focused
on
precision
and
because
of
that
they're
they're
not
able
to
they're
not
comfortable
with
some
of
the
things
that
they
extract
and
when
they
do
that,
what
they
do
is
they
basically
thumb
balance
sheet
and
that's.
D
So
if
you
understand,
if
you're
looking
at
it
for
the
point
of
view
of
does
this
thing
answer
the
question
and
stop
worrying
about
I
got
nine
out
of
the
ten
parts
that
it
was
that
was
in
this
paper,
and
it
missed
one
as
I
said
to
them.
Does
that
one
change
your
opinion
right?
So
it
becomes
more
of
an
engineering
than
anything
else
now.
The
other
thing
is
I.
Think
what
we
we've
done
and
we've
tried
to
say
to
them.
D
C
D
C
D
A
B
D
That
kind
of
effort-
that's
a
great
question.
Actually
what
we're
trying
to
do
is
develop
a
series
of
images
that
combine
the
ingest,
analytic
and
output,
so
that
a
person
who
wants
to
work
in
this
field
will
have
a
package
that,
because
boy
into
their
infrastructure
and
do
the
psychoanalytic,
the
questions
were
developing
it.
Now.
No,
it's
not
ready
yet,
but
we
hope
to
have
it
ready
sometimes,
but.
C
B
C
A
very
focused
interest,
and
why
are
we
so
interested
in
that
is
because
the
more
information
we
have
coming
in
and
the
kinds
of
questions
that
people
have
there's?
No
doubt
there
will
be
a
pattern
that
will
allow
us
to
figure
out
how
to
leverage
the
partnerships
that
we're
trying
to
develop
here
with
the.
B
C
A
If
I
could
give
an
example
from
the
patient
advocacy
side
of
the
value
of
and
Europe
to
your
question
about
the
algorithm
and
are
you
matching
symptoms
to
rare
diseases
and
what
our
patients,
but
also
patient,
advocacy
groups?
Really?
What
are
they
look?
So
I
talked
earlier
about.
Patients
want
to
get
diagnosed,
but
they
also
want
to
participate
in
research
because
hopefully
they
want
to
find
an
effector
or
treatment.
And
so
it's
a
real.
A
Refers
to
the
disease
that
affects
my
family
is
time
on
was
misdiagnosed
with
multiple
sclerosis.
So
when
she
was
diagnosed,
you
know
I
thought
about
it.
On
a
personal
level
that
one
of
my
organizations,
the
initiatives
with
you
know,
we
know
it's
always
misdiagnosed
as
multiple
sclerosis
or
diagnosed
when
you're
older
in
life
as
Alzheimer's
disease,
I
had
a
list
of
all
the
various
15
things
with
you
know
commonly
misdiagnosed
yet,
and
so
we
could,
we
I
usually
said
you
know.
Who
can
we
just
go
to
locals
versus
centers?
A
A
But
really
it
was
just
saying:
hey
have
you
heard
of
you
know
patek
philippe's
heard
of
distances,
and
you
know
what
would
you
test
where
I
went
to
test
with
and
what
we
heard
heard
on
a
small
scale
from
from
ms
doctors
was
oh
no
so
far
low
on
the
on
the
differential
for
multiple
sclerosis.
Multiple
sclerosis
is
kind
of
a
negative
disease.
A
You
can't
definitively
on
diagnosis
that
you
kind
of
exclude
everything
at
cindy's,
so
we
you
know
we're
still
persisting
and
saying
communities
considering
as
something
to
exclude
and
just
really
haven't
gotten
very
positive
feedback,
and
so
what
a
patient
advocacy
organization
has
had
to
do
and
various
other
diseases
similar
to
this.
It
has
to
create
these
patient
profiles.
Themself
the
one
of
my
slides
said
you
know,
and
what
I
approached
another
Medical
Center
saying.
Could
you
do?
Could
you
do
a
study
and
catechol?
They
said.
Oh,
this
is
intriguing,
I,
really
love
to
do
this.
A
C
Gee,
what's
so
fascinating
about
this,
is
that
workaround
solution
there's
causes
frustration
for
that.
The
patient
advocacy
group
is
what
it
takes
change
the
way
we
think
about
diagnosing
disease,
there's
also
about
how
you
educate
people,
because
you
have
a
limited
amount
of
time.
You
have
a
limited
amount
of
resources.
C
B
C
So
the
reason
I
like
dealing
with
human
behavior
is
because
you
don't
want
to
say
one
thing
and
then
do
enough,
so
in
other
words
we're
all
in
this
together,
we've
taken
a
protocol
who
want
to
help
people,
but
then
we
are
uber
protective
of
our
data
that
we
won't
share
or
will
say.
Well
now
we
have
a
collective
IRB
system
which
we
are
trying
to
develop
through
our
clinical
translational
science
award.
But
then
the
IRB
s
will
operate
like
the
County
Sheriff
did
50
years
ago
and
say
not
in
my
County.
D
C
D
To
do
you're
saying:
oh,
it's
all
well
and
good.
We
ways
of
addressing
this
problem
and
we
just
kind
of
perseverance
but
I've
got
to
tell
you
it's
really
remarkable
and,
to
a
certain
extent
understandable,
but
it's
why
we
get
altered,
especially
in
the
error
of
data.
If
we
really
want
jeans
to
help
us
out
and
make
our
lives
easier,
we
got
to
get
over.
C
My
hypothesis
is
that
the
value
of
the
data
sharing
isn't
enough
for
people
to
overcome
the
risk
averse
component
of
it,
and
so
they
they're
able
to
use
that
as
an
argument
that
it's
not
worth
doing.
It's
not
going
to
really
fundamentally
change
the
way
they
do
business.
We're
trying
to
get
people
to
think
using.
Maybe
a
couple
of
examples
that
that
the.
B
Effect
what
you
said
initially,
we
don't
have
time
to
wait
every
day
that
goes
by,
but
just
so
just
speed
up
a
diagnosis
or
you
know
Sharon
King
they
own.
The
tailor.
Still
in
his
reasons
today
is
her
daughter
was
diagnosed.
She
realized
look
I
know
my
daughter's,
not
gonna,
be
saved.
I
will
make
sure
that
other
parents
don't
have
to
feel
this.
Yes,
and
so
every
day
somebody
is
territorial.
Wet
is
to
let
to
a
glandular.
A
B
A
B
C
A
C
C
Transported
first
across
the
south,
these
data
hubs
and
then
across
the
nation,
is
fundamentally
failed
and
and
subjective.
So
what
it,
what
it
must
do
because
of
the
rarity
of
a
disease,
is
in
ischia
graphic
distribution
at
the
gecko.
It
has
to
include
an
answer
to
that
question
and
Michael
understands
that,
but
I'm
gonna
have
to
let
him
explain
how
it's
gonna
get
done.
D
What
we
said
is
that
yeah,
as
I
said
at
the
beginning,
is
we
have
this
process
that
allows
us
to
go
in
and
interrogate
an
organization
how
they
want
to
approach,
and
what
that
does
is
it
focus?
Is
the
activity
hopefully
away
from
some
of
the
perceptions
to
really
what
need
to
answer
questions
now?
The
beauty
of
the
system
that
we're
developing
is
it's
all
open
source
and
we're
designing
the
system
so
that
it
can
be
reproduced
for
business
record
a
little
bit
or
institution.
D
There's
nothing
proprietary
about
this.
You
can
get
it.
You
can
do
your
images,
and
this
is
what
we
hope
to
have
at
the
end
of
the
air.
Is
you
pull
up
an
image
you
want
to
do
the
in
chess
part?
You
wanted
analytic
part,
you
want
to
do
the
display
art,
it's
all,
bundled
together.
So
if
you
think
about
how
we
can
approach
an
organization
and
no,
we
can
talk
about
hierarchies
and
data
and
data
security.
D
Put
you
can
install
all
these
analyticals
behind
your
firewall.
In
your
analysis,
the
remanence
venture
going
to
get
from
us
is
look
at
its
infrastructure
working
right,
that's
a
win-win!
Now
we
have
to
start
looking
at
that.
What
we've
done
is
we've
taken.
Okay,
you
said:
there's
nothing
in
it
for
you,
we've,
given
you
something
you're
saying
you're
concerned
about
security.
It's
inside
you're
in
binary.
You
want.
C
Imagine
it
builds
that
infrastructure
across
country
how
it
would
can't
form
the
way
we
look
at
all
of
these
problems.
Search
air
and
economics
I
mean
it
would
be
an
extraordinarily
valuable,
powerful
tool
that
we
could
all
benefit
from.
It
would
make
our
lives
position
and
exciting
and
vibrant
and
allow
us
to
do
what
we're
here
to
do
better.
B
A
A
Yes,
we'd
like
to
extend
a
warm
thank
you
for
a
panelist,
but
they
have
a
few
announcements
before
we
close
first,
if
you're
interested
in
attending
the
southern
data
science
conference,
which
South
big
data
hub
co-sponsoring
our
early
bird
discounts
and
a
special
south
of
big
data
hub
discount
and
today,
so
the
discount
is
f-ed
hub.
2017
is
the
code
and
you
can
learn
more
about
it.
At
the
step.
Southern
data
science
website.
B
A
Tomorrow
will
host
the
last
webcast
of
the
South
hubs,
vital
data
series,
which
is
run
out
of
Georgia
Tech,
our
partner
in
the
hook
from
1
to
2
p.m.
Eastern
visit
your
account
our
calendar
of
events
on
the
South
hubs
website,
South,
big
data
hubs
org,
and
this
Friday,
our
infrastructure
and
data
sharing
working
group
will
be
needing
from
three
o'clock
to
430.
We've
got
some
great
demonstrations
and
discussions
there.
Another
panel,
both
of
groups,
one
Jennifer
hammock
from
the
encyclopedia
of
life,
will
be
talking
about
for
biodiversity
sciences
and
data
science.
A
We've
got
the
open
science
framework
books
coming
on
friday,
which
are
excited
about
that
again.
Go
to
South
big
data
hub
Oregon,
look
at
our
calendar
of
events.
Finally,
our
next
South
big
data
hub
round
table
will
be
focused
on
Dione
march
nine
from
noon
until
one
thirty
eastern
and
it
will
be
a
panel
on
anti-social
shooting.
So
things
like
vices,
yes,.