To give you a flavor for
the nature of our business and

the in order laxity because if
you're going to use the data.

To simplify our business which
it requires it's important and

context this is a I'm giving you
a one hour race through health care.

Which you know would be the first lecture
in a course on beginning to do a deep dive

and so I'm going to it's going to be
just layers of sort of insights about

what this business is about I'll close
in the last five to seven minutes and

talk about a model that we're pursuing.

Which is intended to really bring two

worlds together that largely act
independently of the world of health care.

That has thin margins has
mountains of data and no resources

to use the data and then you have
a market which has mountains of money and

talent but doesn't know anything about
our business and so they invent things.

That I always say reminds me of a Seinfeld
episode many of you probably don't watch

Seinfeld but if you did there was
a one one show where Kramer is

talking to the C.E.O. of a company who
is reviewing his report and the C.E.O.

says it's almost like you don't know
anything about our business and Congress

as well of course I don't I'm volunteered
here I'm a volunteer worker I don't

know anything about it I'm just starting
out and that's what if you see a lot of

inventions that come our way that's what
it feels like which business is for so

my intention is to give you a flavor for
what this business is about and

I like to call this lecture doing
the impossible because these things

of making care more affordable more
accessible waste free high quality

care and high quality experience are the
things that we wished we could do and

we've been wishing this for
I'm going to say at least five decades.

And we don't get there we there's
always it always seems that.

It's a distant vision away and

I think we don't get there because there
are structural challenges with the way in

which we deliver care that will
never allow us to get there

given the way that we operate today
because we're still very much twentieth

century even though we don't know
what's on the health record it's still

a twentieth century construct that
drives how we actually deliver care.

So I'm going to start with things
that you should know now if I

gave a talk on things you
should know we'd be here for

the next two years I spent my
first ten years at Guy's Inger.

Learning what the business was about and
it was an ordinary It was an extraordinary

complex journey to really understand
all the facets of the business

because it's a really complicated business
health care is not a business it's

a hundred businesses and each of those
businesses has dozens of problems and

that should reveal to you how challenging
it is actually to transform this business

because it's not just one thing there is
no silver bullet there is no one model

that will transform it in a way
that would make any sense.

There's lots of problems to solve so I'm
going to go through just some things that

are important to know it's not everything
so I want to start with how how markets

behave because health care to some to be
teaching is the same quality where we

have knowledge workers it has this sort of
quality to it which is when a market first

starts out it's a new market what
happens is that it expands what

that means is that it takes
a growing percentage of the G.D.P.

right it accounts for Grow a larger and
larger share of the G.D.P. and

then competition sets in and
things get better faster and cheaper and

then over time the share of the G.D.P.
that that sector accounts for

shrinks because other sectors
are growing and things of that sort and

eventually it just gets very very
efficient that's how most markets work

that is not how health care works right
health care is a completely different.

Animal And so I asked the question
when you think about the total health

care spend in any country what
is it most strongly related

to G.D.P..

And there's a reason why and I think the
reason why has something to do with our

brain stem is wired you may say what
the heck are you talking about.

So if you look at this wonderful
study that was done by the W.H.O.

they took data from one hundred
seventy four countries and

they looked at the per capita spend the
total health care expenditure per capita

in relation to the G.D.P.
per capita it's on a log log scale.

With R.

squared of point nine five right so

what that tells you is something really
fascinating about who we are as humans

what it says is that as
the wealth of individuals goes up

not only will we spend more
of our wealth on health we

will spend a greater share of our
wealth on health I think the reason why

that is the case is because we've run out
of the things that are meaningful for

us to spend things on as we get wealthier
especially the super wealthy right and

the only thing that actually matters at
the end of the day as you get older is

your health I'd rather invest
in that than buy another car so

this model really states that
the very explicit way it's wired

into who we are so there's something
about this dynamic between

how we think about longevity and
surviving and the passion to survive.

And the way in which markets
unfold that play off of that

that motivate more spending right
because it's not just us but

it's also the way in which the market
evolves so it's very very hard to compare

for example the U.S. to European countries
because the G.D.P. per capita is different

you have to if you don't take into account
the G.D.P. it's a kind of false comparison

our labor costs are higher in the lots of
things are higher here than elsewhere.

Now there's lots of nuances to this
because some of it has to do with

practiced behavior some of it has
to do with fee for service versus.

You know verses contracted care or
things that that are nuance but

the relationship speaks volumes
in terms of what it is that

drives individuals and markets to
behave in this way it's not it's not

anything that's specific to any economy
it's specific to the nature of economies

and the questions about that

that surprise you you have that look and
so I'm going to.

This.

Here.

Now is.

I'm trying to log logs not linear so
you're got to you've got to account for

that right so you start to really
accelerate as you get wealthier

now the model would predict that
our G.D.P. our share of G.D.P.

should go down as as
the average per capita income

goes down right we're starting to
see that the growth rate in the.

In the cost of health care is actually
starting to go down quite a bit

although it hasn't plateaued and
it's predicted go back up one of.

It's it's in there all their hundred
seventy four and you're up there and

you're right.

No Well this is this is two thousand and
five so this is probably before they

they modernized but they have a they
have a quirk in G.D.P. right now

but if it's this it's the way in which the
country's lined up on a log log you don't

only spend more you spend a greater share
so that's why it can percent of G.D.P.

in the U.S. is in part the way it is it's
also a peculiar part of our economy and

the way in which health care is structured
right it's somewhat It's not unique

to the U.S. certainly Germany as an aspect
of it the U.K. even has an aspect of this

sort of commercial side of our business
but that doesn't drive it alone.

Where our G.D.P. is pretty high yeah
log log it's a log relationship

Well I'm saying if I was if it was on
a linear scale it would look completely

different you'd see that outlier but
when you when you when

you transform the scales it tells
you something about the behavior and

it makes sense as we start to get really
wealthy What are you going to spend it on

in the last five years you know because
that's where most of it's going.

To actually.

Yes I know.

So just a few things that you
may already know this right but

when when people talk about health care.

You have to ask yourself well
what part of health care

because health care actually
has lots of different.

Sectors and I've only listened a few here
and I've done something else I've also

listed what that health care is dominated
by who's the major player right so

we often think that physicians
are the dominant player everywhere but

they're not physicians are dominant in the
setting they are the player around which

everything gets built how many of
you ever been and then patient.

So how often did you see the nurses
versus the doctors it's a nurse business

the nurses make everything happen and
if there are mistakes

the nurses also make those happen
right the doctor sort of comes in and

out the doctor make me do surgical
procedures and things of that sort but

it's a nurse nominated business and
so when you think about

the ways in which big data could come in
and you're looking at the inpatient side.

You've got to think about who's the player
that's driving most of the care and

how are they doing it what are they doing.

So this is just a you know a short list
you can dissect this even further and

I'll come back and sort of talk
about that but part of my reason for

sharing this with you is to
understand the complexity.

And then we have this thing called
coordinated care so this is an idea that

was invented and
started to emerge as health care and

the seventy's and in particular
in the eighty's really started

to explode after Medicare Medicaid
legislation was passed under the Johnson

administration around one hundred
sixty eight and so this changed and

transformed the business in a way that
was profound and we started to have

this problem called fragmented care people
were getting a starting to get older and.

And most care involves more than one step
that you have to connect to things and

nobody was thinking about the roadmap
that the customer needs to know that

you've got to get this done first and then
this done second I missed on third and

so coordinated care came into existence.

With the recognition that those handoffs

don't get managed very well that we
mostly leave it to the patient but

I will tell you that this concept which
has been around for a long time and

has been organized and represented
them deployed in an amazing number

of ways is a failed construct it
doesn't really work that well and

it follows a twentieth century model and
that model says that if we have a problem.

Let's throw more labor at it because
certainly if we do that we can solve

the problem but the problem with a
coordinated care construct is the problem

that we have with lots of ways in which we
try to help patients as customers which is

that you put a human in a place to
do things in an unstructured way and

try to make things work and
they form relationships with patients and

then that person moves on and
turns over and there's no memory

of procedures connections
relations things of that sort and

it gets chaotic for a while until somebody
else comes in and that kind of process is

not sustainable over time that it's
just a very very fragile process and

the notion that we think that
we could manage patients and

coordinate their care is just a failed
notion in many ways I believe we

should be reversing it that we should be
equipping the patient with a navigator

that ensures that I know exactly what
talks to my record it knows to inform me

exactly where to go and it gets my team to
make sure that they get things done on my

behalf at the right time and at the right
place we've got it wrong navigation

is a construct that we know on the day in
the digital side how to do well it's just

that nobody who has the skills that
you have has invented that yet but

I would tell you that if navigators
were built for oncology or

other areas it would transform
the business in a way that would be.

Would go back to those
sort of basic games so

navigation in my view is a important
twenty first construct that really

should get rid of this idea that
is largely a failed construct.

So the other thing that recognises
that we have lots of data lots of

different categories of data we don't just
have electronic health record data even

electronic health record data
there's there are diverse types and

I haven't even listed.

All of those diverse types you can you
can talk to Ed which I who can tell

you a lot more about these data some of
these data are really interesting right so

we often think of just the structured and
unstructured clinical data.

As the most important source of data but

if you really wanted to be on the cutting
edge I would say look at the activity log

the activity log data is the data that
every electronic health record is

required to have to track anything that
you do in the record right it's a security

data set it's intended to make sure that
nefarious things aren't happening or

that the record isn't being used by
somebody you shouldn't be using it

in those data and
my view are a incredible resource for

beginning to think about how we
develop anticipation engines so

if you're a doctor over the course
of a year you have millions of

activity long records you could connect
those to why the patient was there what

happen to the patient you could begin to
imagine how sophisticated analysis of

those data could anticipate what
a doctor wants to see given the dish and

on the patient and why they're there and
you could really begin to transfer think

about you could transform the ability of
that doctor to get a lot more done and

a lot more satisfying ways
even constructing progress so

that database is there are in
the literature right now maybe

six publications that
abuse that database and

I think chemist probably will have two
of those you'll ramp it up to eight but

it's a unique resource very very important
there's other data that's not here that's

coming in and one of the interesting
things that's changing in this world is.

It's how much new data is coming into
everybody's electronic health record there

used to be this idea called health
information exchange we had in the and

after two thousand this idea that we could
we could build consortiums provider and

insurance groups and
we would all put our data together and

share a failed construct There was no
business motive behind it there was no way

to really sustain it we went through
three generations of that concept most.

Funded by the federal
government all failed

what it's being replaced by is the market.

Moving data around so

there's a company called Sure scripts
sure scripts is the company that's

the back end to all pharmacies
that does all of the adjudication

your doctor writes you a prescription
you take it to the pharmacy pharmacy.

The pharmacist gives you medication a
record is transacted by sure scripts that

said you picked it up that record now
goes to the place where you got your

rights is now sitting there parked
in a database that nobody can use or

nobody does use because
it's cumbersome data and

in general just throwing more data in
the record doesn't really help unless

you manufacture tools to automate
the visualization of it right but

one thing that's important just
about that data alone is that

if you're on the electronic health
records side you only see what's ordered

we know twenty to fifty percent of
people never pick up their drug right

if I as a doctor knew that and I could
manage that information effectively and

automatically I could do a lot more in
trying to get more patients on board and

have them overcome some of
the challenges that they have and

taking medications some of it's just
belief driven I take it for two weeks and

then I'm good so people don't understand
often that this is a longer term play that

will require more adherence So when you
think of data it's important to recognize

that the kinds of data that are growing
in records is diversifying in

substantial ways the other thing that
I think is important understand is

what is healthcare and what is it not
health care is not wellness we are not

in the wellness business I can't tell you
the number of times I've been on panels.

When I was with the guys in your and at
Sutter and people would say What are you

doing about wellness and I would say well
nothing because that's not really why

we're in the business people don't come to
us for wellness there's much better places

to go people come to us because they have
a problem and they want to solve it people

actually don't like coming to us how many
of you love going to see the doctor.

So not many it's not something
that we actually try to avoid it.

So we solve problems that's our business
and we don't do wellness and so

if you're thinking of inventing something
for wellness there's a different pitch and

a different customer and
a different person who's going to pay for

it than if you're thinking about inventing
something for health care problems right.

That's very very different domain
always important to follow

the money because if you're going to
invent something that you want to get to

the market you have to know where the
money is going this is a two thousand and

ten chart and I'll just point out you
know where the interesting changes are so

everybody's trying to keep people out of
here out of the inpatient side right and

we're starting to see you know
fairly substantial impact

in and a lot of local markets so
that's that's interesting we

the bed capacity that we have nationwide
is substantially more than we have and

that trajectory is likely to continue
even as the population ages because

of the ways in which we're
moving surgery to amatory and

the ways in which we're really working
hard on reducing length of stay and

trying to better manage patients under
a contracted arrangement where we're

trying to keep them stable especially
those who have chronic diseases so

we're likely to see this shrink
this probably grow to some degree

this is likely to grow substantially
at the prescription spaces where payers

are very very worried because
we're in this special era where.

Genomic.

For the first time really seems
to be translating into meaningful

new generations of therapeutic and
they are really quite extraordinary but

they're also extraordinarily expensive and
so this is a big worrisome area

in terms of beginning to mushroom the
costs of care because the therapeutics for

chronic diseases for diseases like M.S.

are fifty seventy thousand dollars
a year the drugs are fantastic.

But these costs are really quite
substantial So we're likely to see those

kinds of trade offs I think between those
two sectors but it's important to be aware

of that and I'll just come back to
this this is what we care about.

The sort of never ending dream
that we want to achieve and

then I will say the biggest deal ever
in health care is anybody have a guess.

Where the biggest potential win is

waste I'm not talking about
industrial waste I'm talking about.

Waste that care and

you know the evidence is pretty consistent
that about a third of the three

trillion plus dollars that we spent last
year is just waste if we didn't do it.

Nobody would be better off
except those who get revenue for

that Karen are dependent upon it so
this is a really important target because

it's a kind of win win but the problem
with this is it's in every nook and

cranny of healthcare it's not just.

You know the waste is in
a vault someplace and

you've got to get it out you actually have
to develop sophisticated tools to surface

it and that's just the first step because
surfacing it doesn't actually get

rid of it you have to actually then have
a way of mitigating it or minimizing it so

this is where there is
huge opportunity where.

Affordability could go up.

In many ways experience of the patient
to be better because unnecessary care

is a net plus and where sort of everybody

wins letters.

And policy you know challenge
how to address the reason

actually United Way not clear why
the very extensive literature

I think what we haven't done
enough is to take a fairly

substantial body of evidence on all the
different areas where waste is and really

do a much more precise job taking large
data sets like a set of data set and

explicating all the different
areas where it is and

subjecting it to the kind of analysis
that we ordinarily wouldn't the business

how hard is it to get rid of it how
much is it you know that usual kind

of analysis where we can begin to
prioritize where the winds are.

But the challenge the other part
of the challenge is that that and

we may never get to this I think unless
we have more sophisticated tools is that.

I think there was a lot of belief a decade
ago that we're going to move increasingly

towards managed care you know contracted
care where where my payment as a doctor

was about the quality of the care that I
delivered not about not about just fee for

service but
I don't see any evidence that we're going

to be moving away from a kind of mixed
economy around health care we're always

going to have sort of a managed part in a
fee for service part is always going to be

sort of the norm there's just the markets
across this country are too diverse and

they're never They're never always
staying one way or the other and so

this is a this is a bit challenging
I think because of that so.

The the most remarkable thing
I think that's changing.

In health care is the growth and

medical knowledge this was the product of
the first revolution in health care which

started with flexion or who reinvented
medical schools in two thousand and

nine hundred ten right where at
the time he wrote a scathing review and

I don't know forty or fifty of the schools
in the US in Canada were closed down and

sort of rebuilt on the on
the Hopkins model but

if it went with that effort led to was
the reinvention of physicians training and

then eventually nurse training in
the training of other providers.

And this bed to bedside construct which
was part of the original idea in the US

back in one nine hundred ten and
the most extraordinary

knowledge engine ever invented
extraordinary mind boggling So if

you graduated from medical school in one
thousand nine hundred fifty the doubling

time of knowledge medical knowledge
nine hundred fifty was fifty's so if you

graduated in one thousand nine hundred you
were pretty much in authority for the rest

of your career if you graduated in one
thousand Can I mean two thousand and

ten the doubling time was three and a half
years if you graduate in two thousand and

twenty just two years from now so

you start two thousand and twenty and
you go to two thousand and twenty one

the doubling time of knowledge medical
knowledge with three and a half months.

So knowledge is exploding in a way that.

Is impossible by traditional math methods
to keep up with it's not just data but

knowledge is exploding we've had this
problem for thirty years right knowledge

has been growing so fast that that our
ability to translate it into practice.

Has been exceeded that dominant way in
which we translate knowledge to practice

still is by graduating
a new coder physician.

Because what happens with physicians is
they are at the peak of their knowledge

because they can devote time to stay
knowledgeable when they graduate and

then it's a struggle to practice and
stay knowledgeable for

most physicians over time and
so we just yes well.

There this is I'm just
taking these estimates from

articles in the literature that
I've sort of done that kind of.

Well I think I think one way you could
do it is you know one way you could

translate it is what gets into evidence
and so if you look at guidelines exploded

I mean there's there's formal knowledge
there's knowledge is published and

there's different ways of reading
the quality of the of the knowledge itself

and I was three forty five years

we didn't know anything and
we really did in fact had a professor who

famously professor who pointed
out that if you have a patient

with a problem less that surgically
treatable or treatable with

antibiotics cure wasn't really what
we could do that was forty five years

this is what he said today we're using the
sophisticated complex you know we derive

medications actually do work now and
change the disease yeah.

And that's why it's such just explode but
it's just a benchmark right so

there's that it's the trajectory
it's been going on for

decades it's not anything new and
but it's something to be aware of

that there are challenges not just with
data and not just representing data or

when I was trying there were no
there was always that's right yeah.

I'm one of the very first
cat was it really is trying.

Now and cry when I was three that.

You can only go look at it you can use it.

I mean this is this is in my lifetime you
know with a little more than half of my

life that's what I represent but this is
a point but the trajectory started in one

thousand town so one way you can have big
numbers of course is the Stars moment yes.

But it speaks to the incredible
machinery that is you know across.

Hundreds several hundred academic
medical centers universities

industries we are a knowledge generating.

Enterprise in this country and it's that
model has spread we have led that model

and it has been widely adopted
the most widely adopted

knowledge generating model is
the US model it's proving the case

exactly and so the challenge

is not just representing data and using
it the challenge is combining it with

knowledge right that's really a challenge
we have to sort of solve and then one

thing that hasn't changed is the amount of
time that patients spend with doctors so

you have an explosion so data is
doubling almost every two years and

that's that's also on a accelerating
trajectory and that's going to but

what hasn't changed is how much time on
average people spend with a doctor which

is about an hour a year and that's across
all age groups obviously it's it's not

normally distributed it's a kind of
you know skewed left distribution.

With a long tail but the fact that
it hasn't change speaks volumes for

what we have to do to close the gap
between the kind of data that's available

that nobody has time to look at
the volumes of detail knowledge that's

available that nobody has time to learn
and the and the different ways in which we

have to think about how to represent this
in a way that makes it useful it really

does speak to the need to anticipate
in the moment what should show up to

make a doctor a better decision maker
doctors in my view the way that the world

is structured today do things that we'll
look back on and scratch our heads and

think what were we thinking about
you mean a doctor asked patients

opening questions on why they were there
they didn't have all that data front

loaded that the patients spent ten minutes
informing them every detail overt and

covert reasons why they were there I mean
doctors actually spent time looking for

data in the record that's a ten
dollar an hour task that's not a two

hundred to three hundred on our task so

the way that our business is
set up today is that we have

doctors doing things that are remarkable.

Given that they are probably
the most expensive.

Part of the educated workforce and
we spend more on educating doctors and

we do on any other type kind of talent
that we have and then we subject them

to an environment that has them new ten
dollar an hour tasks that's why we're

still in twentieth century restricts
still living in that world so

the business is chaotic and
your job is to introduce some.

Control over the chaos and in my view I'm
going to just walk through a simple sort

of conceptualization of how at
least our team thinks about.

The the kinds of things that you
might want to build in that sort

of in sort of three domains
connecting customizing and

communicating because in our view
these as a source of things that.

That we need to think about.

And and I'm going to describe
sort of a framework because

the challenges that I was describing
to you are not a new story right

they are the oldest story in every economy
they've been around for hundreds of years

and the way in which we solve
the problem is we automate we take tasks

out of the process that human doesn't
need to do you preserve the human for

the higher and tasks where they have
special skills so automation really is

an old story it just happens to be a
little more complicated to figure out how

we would actually do it in our business
because as all walked through what health

care is about and that sort of inner
core it is an orderly complicated and

it's inordinately difficult to think
about how you insert analytics and

other things into a process that
that is in itself complicated so

when you think about
the elemental components of our

business in many ways it does
come down to data knowledge and

people and there's lots of different
ways that you can put these together so

these are the elements the dominant
elements there may be others but

there's all kinds of combinations
that actually go on in healthcare

that are represented by.

What I'm showing here.

Now we do a lot of this manually right.

There's data to data communications
that the doctor might do

the doctor might look at a blood
pressure measure of a patient and

then bring up what they've been treated on
and if they could bring up whether they

actually picked up the medication and the
blood pressures out of control they might

have an insight into recognizing that
this patient isn't hearing and that's

why the blood pressure that's a dated to
data communication and very manual one.

We can begin to think about how we combine
these different elements to represent

intuitive concepts concepts and actually
get to things that we can transact So

a knowledge to data connection
is what I would call a care

gap guidelines say that
if you have hypertension.

And it's been this long that you're out of
control in this is what I should do and

if I haven't done it then there's a gap
in the care that I'm providing and

I could I could automate the process
of both the technique that and

signalling to try to get that
care gap closed right so

there's ways in which we want to combine
these different components that add

meaning a data to knowledge is
decision support if I have data and

I evaluate it with knowledge I can use
that combination to begin to think about

recommendations so the building blocks for
how we use data and

how we analyze it begins at this sort
of elemental level treatment adherence

I have my order data I have the sure
scripts data I recognize that

you're not a hearing so
one of the things that I

always wonder about is whether or not
the machines can make these connections.

In my experience I would say that.

That there is this blend of scalable.

Expert knowledge that when combined
with the machines are capable of it

that's the best combination.

If you took some of the work that we did
with ontologies like medication ontologies

and disease coding ontologies we realized
that when we compress the dataset and

the feature set using those ontologies
two things happened the model was more

parsimonious so it had fewer more robust
features and it predicted better so

that's a good thing because I get I get
a lot more out of it my guess is that

we would do well to begin to think about
how we translate the different elements

into intuitive constructs that mean
something to us because my guess is

that it would improve model performance
and model interpret interpretability

the next thing that you have to think
about is all right so what if you have.

These constructs now that
you've created in your database

you have to do something with
them because if all you do is to.

Go through all of Sutter's data and
in real time convert those data into these

higher level concepts somebody still
has to find them somebody still has to

know that they're there nobody has time
for that and that's the dilemma right so

you have to find a way to surface it and
that's to be automated you have

to integrate it with a workflow because
one of the things we learned early on

with electronic health records and alerts
is alerts just inserted themselves into

an encounter in a way that made no sense
to practitioners right there on a card in

the process trying to get something down
all of a sudden an alert shows up ask them

a question that's out of context because
that was somebody else's job trying

to insert their job into a physician's job
so you have to sort of think about how to

integrated orchestrate it in a way that
makes sense to the practice context and

then you have to actually make it happen
you have to actually visualize it

in a way and I'll come back to that one
thing I would strongly recommend that you.

Read is the literature on jobs there

are two leaders who have really written
this literature one is you'll wake.

And Christians and they used to work
together and then something happened and

they went their separate ways but they
they both gave great insight to the notion

of jobs and if you're going to invent
automated tools if you're going to invent

ways to transform a business you have to
understand what it is that people do and

a job isn't I'm a doctor a job is
something much more refined and much more

and there's dimensions to a job so if you
read your books work he would say that

there's a functional element to a job
there's an emotional element and

there's a social element and one of the
things we often do is we just think that

the only important part is the functional
and we leave off those other two

dimensions and often nobody is interested
in our invention because it wasn't

relevant to how they felt about it but
the notion of a job is.

I go to a store going to a store
to buy something is a job right

I'm trying to fulfill a need and
there's a functional social and

emotional component to that So
this is a very important construct

that I think is missing from health care
and I was actually saying to demanding one

thing that would be interesting to pursue
is funding to build the jobs ontology for

health care because if we
had a jobs ontology for

health care it would really begin to
transform our ability to map data to.

That ontology in a way that allows us to
begin to scale applications that transform

the Sorry I should have
gone through that and

so just to give a sense of
what I mean by a job so

if you look at a primary care
physician during and counter

these are the possible jobs that they
would accomplish during that encounter.

And.

Every one of these jobs
has those three dimensions

every one of these jobs has a task list
related to it right so it's actually

much more complicated than than what
is represented in the simple diagram.

And let me just sort of then sort of walk
you through a quick scan of a ontology and

my intention is to give you a flavor for
the inordinate complexity of our business

right to really recognize that
the reason why we can't deliver on those

quintile AMS is we have too much to do
there is just simply too much to do.

And the business is too complex and
so this is the high level ontology

the setting of care whether that specialty
walks down in a way that kind of fits and

ontology and so these are the different
settings I didn't capture all of them but

you can take one of those settings ambles
door and you can say is a primary care or

is it specialty care and
I can take one of those neurology and

I can say well what kind of transaction
of my having is it is it is I'm and I

looking at the episode and the episode is
a constellation of two or more encounters

that go together in a way that are related
to me managing somebody has problems is it

at the encounter level am I looking within
the encounter at the job of the task or

the connections that I'm trying to make
from the data to get things done and

then I can take that episode and say
well wait where am I in that episode and

I could say well it's the post is that
encounter and what am i actually doing and

then within that even an encounter I can

say you know in the visit these are the
things that I want to accomplish I can

take one of those jobs and say these
are the tasks related to those jobs so

when you break this down when you take
every job that somebody has to do and

break it down to its tasks and then go
to the data elements something becomes

very obvious there is a obvious reason
why we actually can't get to the quintile

lands is an obvious reason why doctors and
others can't get their work fifteen or

we need machines to
facilitate the process.

So the last step in building capabilities
in building tools is you have to

communicate right so you can connect
things and build intuitively sensible

constructs you can analyze those to reveal
insights you can customize things in a way

where you're bringing it all together
to fit into the workflow but then.

At some point it has to be transacted
in exactly the right way at the right

time that fits the cognitive process
of the provider at that moment and

is meaningful at that moment right so
that that requires a recognition of

a simple representation of it a clear
representation and one that actually

captures the emotional content that's
relevant to that moment there's that sort

of empathy that's represented in it so
when I think of health care solutions.

These are really important elements
that are sort of my targets right

it has to create value right otherwise
nobody's probably going to buy it.

Somebody might be excited by it but
they nobody will pay for

it it has to simplify work it
really has to connect here

because otherwise it may not
resonate with somebody so

alerts just show up never resonated with
anybody because they weren't actually

helping anybody get the job done except
the person who designed the alert right so

that person's job was getting done.

And then it has to connect care crossing
counters because that's a really important

challenge now when you think of
the process and this is a snapshot.

Of what you actually have to go through.

To bring something to market
that's really successful

has to start with a great idea and
when that evolves over time.

And you have to have to convince yourself
and others that you create value for

something now often what happens
is that you're creating value for

one person with somebody else has to pay
for it so that's made the job a little

more complicated because you have to
excite the one that you want to use it but

then you also have to excite the one that
you want to pay for but the way you're

exciting them is going to be completely
different so it's a different kind of sell

it's often easier if you're going to have
the one that's using it and that's paying

for it in the same person sometimes in
healthcare that's really hard to do and

at some point you need real data
to validate your idea not to build

the solution but you have to sort of make
sure that you don't have an idea that

has nothing to do with health care if to
make sure that you don't end up in that

scenario I described where Cramer was
talking to the C.E.O. and the C.E.O. said

I don't know what your what if you know
you know nothing about our business so

you got to use real data to
either correct your perception

of what it is you think you can
accomplish or to give shape and

context in details to it that's
sort of the validation step.

Then you've got to convince somebody else.

Of it and then and then you've got to
develop a prototype with real data so

now you're using the data to
develop a tool to simulate it

pilot test it to do all that really hard
work and in the real world start ups.

Are given eighteen months to go from here

to here and work with a few systems or
convince a few.

Provider groups to test their application
and that's an eighteen month journey and

most never get there because they
can't get access to the data.

They can't get access to enough data and
they die on the vine because that journey

is is too hard if you've ever tried to
get data out of people in my business but

then even if you do get to
hear the the the real step

is testing your application
in vivo because if

I'm a funder before I give you the next
eighteen months of funding I actually

want to see proof that it actually worked
before that somebody was using it and they

liked it I don't expect you to sell they
just have to sort of adopt it right so

that's that's a very very difficult
journey and the failure happens.

Very very often and so
there's that part of it and then you have

to recognize in this business who it is
that you're designing this for Who are you

creating value for right is it the patient
if it's the patient then you've got to

figure within who is paying for it is it
the insurers the patient paying for it and

if it's the patient then which which
segment of the U.S. or other marketing

are going after because if people
are paying for it themselves and they have

very very high expectations right if it's
the insurer that maybe the insure sees

value in what you have to work out that
value proposition to really make it work.

And then the.

Final challenge is how do you
actually get into healthcare right

imagine that you have the absolute
best predictive model in the world

that could pick up
Heart Failure to use your now.

With ninety five percent accuracy and
that could actually inform

the physician about exactly what to do
to give more life to a patient and it's

all in your black box and so the question
is well where do you take that black box

because health care is everywhere and
where you installing it right so

even if you have something that's really
fantastic the big challenge is how are you

going to actually get it into a place
that can actually use it and it's

a really really hard problem to solve so
when you think of all of this it's a it's

not difficult to realize why we're still
in the twentieth century we have we have.

A challenge to you know really fundamental
challenges to sort of overcome you so

in the last year I've been
thinking about this idea.

Of an Institute for
Healthcare analytics as a nonprofit and

the idea didn't come from me it came from
colleagues of mine who are entrepreneurs

and in Silicon Valley and have been in the
analytic space for years and have lived

that journey successfully of doing
startups and bringing things to market.

And they've been bugging me for
a while to form an institute

that brings these two worlds
together that really should be

interdependent are really interdependent
but not in the very functional way

they're acting independent of each other
and not having much success right so

we have the nonprofit health care
delivery side of the world and

we have the for profit invention
investor side of the world right and

this side of the world is trying to
influence this side of the world but

they have all that friction
in trying to get there

because it's not the investors who
are going to invent things they pay for

things but they face a lot of failure so
if you look at those

two sides of the world you say well
what are their assets so health care has

places where care is delivered that's good
because that's where the customers go.

They have lots of data.

Profoundly under-utilized they have
a deep understanding of problems.

Really deep understanding of problems.

And they have a place to test ideas
in reality and even in the market

has from a healthcare perspective
infinite money because in healthcare

our margins are really really thin when we
look to the investor market we say well

they have infinite amounts of money and
from our perspective in the analytics and

digital space they have infinite amounts
of talent because we would never think

to hire that kind of talent as they
probably would never come to our space and

if we did get them to come to our space
we're not sure what we would do with them

because we're not sure that we know how to
manage them because that's not really our

business our business is not reinventing
healthcare our business is actually to

deliver care your business is
to reinvent health care and

to help me so so and and

they are inherently risk takers this
side of the world only gets the people

who are who get high on taking risks
this side draws the risk averse people.

And they are out of the box
they are in the box so.

And then if you look at the limitations

not much net money will never
spend enough to use their data.

And the fact of way will never take
risks no time or wherewithal to invent.

Limited access to data limited
understanding of health care and

solutions limited access to pilot testing
anybody who looks at this would say

well that's this is pretty easy they're
kind of mirror images of each other maybe

we should just kind of bring them together
in a way that's more logical incoherent so

that we can create when winds and
how you know how how should we

sort of structure this and so that's the
idea behind the institute that our vision.

Is to bring together five to six
health systems we don't really need

more than five to six houses because once
you get five to six how systems you've got

seven eight nine million records
you don't really need that many

records you really need.

Enough diversity in the data so
that concepts like blood pressure or

L.D.L. you have all given a variation
in how they're represented so

you can sort of solve that multi
system kind of problem and

where you can see nuances of systems
what we want to just engage five or

six house distance to be partners in
this and we want to take advantage

of of the market and my argument health
systems is that you will never ever

ever spend enough to extract the value
from your data that will never happen but

you could manage markets and motivate them
to extract value from them to help you

with your data analysis problem because
that's where all the talent is so

there's a very logical way in which we
can structure the relationship between

what we do with partners in the institute
and how we inform the market and

in many ways that's that's
the design of this is that.

It's another lecture to walk you through
how we are going to structure this

with the partners but in many ways we can
work with partners to give guidance to

the market right what startups need is not
only an idea of the big problems to solve

and even the modest size problems to solve
but a deep deep understanding at the data

level on the nature of the problems
to be solved if inventors have

that kind of detailed knowledge
they will invent effectively right

because that's how their minds
work they love problems it just so

happens that they get bad advice on
the problems to solve if they had

really really high quality advice from the
people who care then the the health care

side of the business is beginning to guide
the market on where the opportunities

are the critical role though that the
health system can play is to make their

data available so you can get through
those first two vetting steps because

that's where most startups die on the vine
right they spend eighteen months they

never get there if the data were available
you can get that done in six months and

now you're moving on so this really starts
to accelerate in investment market that

would find value in this and then of
course this part here of actually testing

in vivo is where the real critical
value comes where where a start up

can actually get through that first phase
very successful in a very consistent way.

So the way we're thinking of getting
started is we have to have a compelling

offer to the partners right five or
six health systems that we want to come to

the table and
the way that we've structured our pitch.

We're all always selling you know no
matter where you go in life you will sell

right you have to that's part of what life
is about in part is to say look we will

have a chair one team that we will pay for
you will pay an annual membership fee to

leverage that team we will work together
five systems in our leadership team and

in identifying the kinds of problems
that we want to solve right and

we want to walk through a process and
develop organizational and cultural muscle

around what it means to be data driven
if you look at health care hardly

anything is driven by data we do things
the old fashioned way we make strategic

goals reduce thirty day admission we
get a dashboard that shows for every

hospital with a thirty day readmission
isn't when the needle is in the red we say

to the C.E.O. of the hospital move it back
into the green there's no causal analysis

there's no data driven kind of process
at points exactly to why it went

in the red and reveals exactly how you
get it back in the green that is not our

business right now it's not like a
manufacturing process but that is how our

business should be right data driven
means that we walk through any evolve

a way of exercising this so it becomes
the norm that's what we want to spread and

the applications that the market
develops really could be key to this so

if you look in any health system in
the US Even the best even a guy zinger.

You have senior leaders senior vice
president articulating strategies

this year we are going to reduce
that to day readmission and

get to the ninetieth percentile in the
country this year we're going to get our

satisfaction scores up to
the ninetieth percentile in the U.S.

right so there are some things that
are professed at the executive level

the vice president level then takes
that across an enterprise and

they start to translate it into
something that they think corresponds to

what the senior executive said but
nobody is using data to validate that

this made sense here and nobody is using
data to take this to the next level and

that same process of
translating from S.V.P. to V.P.

to director to local manager all the way
down to the front line there is no data

driven kind of process that ensures
that we're all on the same page and

that connects what happens
on the front line and

asks Is it connected and are we doing
our job with regard to what the senior

leadership writes a data driven
actually means that we construct

a model that clarifies the level of
analysis and connections that we make

each of these levels that has
not been invented in healthcare

it's all analytics it's all about.

The world and how we use data
affectively I think that's my last line

I'll stop there OK Thank you.

That was great.

The problem.

I mean it's great it's
trying to get work with.

Your market yes.

So so the rule is going to be that we
will not take two from the same market

not even close to the same market because
we know that the market's aggregating so

and it's a big country fortunately so
but what's more important

is the culture and the commitment so
there has to be leadership commitment to

this idea of being data driven
because it's a it's a tall order.

And there has to be a commitment
to innovating So there

you know there's a smaller denominator
I think that sort of fits that.

And you know it's funny I gave a pitch
yesterday to Al system and I said you know

the members of the annual membership fees
is only about three hundred fifty thousand

and they said really that's all because
they're looking at thinking that they're

going to get ten team members which they
will and be able to work with for their

systems and drive this that that's a like
a no brainer now so the opportunities I

think are substantial even having we were
talking about even having a university.

You know connection to this to
me is really important bringing.

In turns I mean my success with
interns from here has been you know

such a delight that the opportunity
to do this across multiple

data sets to me could be could
take us to the next level.

As you said you know we're going

to have you here.

And what we're trying to do.

How much have you seen.

As you think success and so

how does this go some ways we
don't we don't focus on waste.

So why thirty day readmission why length
of stay patient satisfaction because we

get penalized or scented and so
to some degree our behavior is motivated

by the incentives or the stick that we
get with writes a thirty day readmission

you're going to get beat up if you're if
you have a high paying for it yourself.

But but in the dashboard construct you
know if you go into any health system and

you ask the leadership see
their dashboards phenomenal

dashboards right I mean lots of just
like it's like dials on a factory but

the difference between the dials in
the factory and the dashboard and

health care system if you said to somebody
what's the causal analysis you've done to

know how to move this needle however you
want to move it there nobody uses data and

they use data to understand
where the needle is and

then they know who to go to when the
needles in the red but there's no model

there's no analytic model that says
If you have a thirty day readmission.

Dashboard Here's our application that will
do all the cars on Alice's four it'll tell

you exactly what you need to push to
get them to look in the green that

that product does not exist it should
it should exist but it does so so

part of what else systems do is that their
margins are so thin that they have to

focus on you know where they can
minimize getting hurt and losing money

it's not on bigger vision they don't
have the bandwidth they don't have the.

They haven't been offered a way to change
their business in a way where they can

free up a band.

It's a challenge it's a real
challenge of the questions.

And.

Of course.

It was with a lot of players.

These days largely as a cooler as a child

who cares do you think it is there for
the.

Or if they can what are the same
here I guess is it has a presence

I like to use the analogy of the history
of investment engine omic since the mid

eighty's right so if you look at that
business the one thing that was amazingly

true is how much we underestimated
the complexity of all bikes.

Before we could actually unlock it and
use it to develop better diagnostics and

better treatment it took us five
generations of business development and

collapse before in
the most recent phase it

now really appears that we're doing things
revolutionary at the therapeutics that

are coming out now are mind boggling and
we're just beginning of that revolution so

when you don't know exactly what journey
you're on you have a simple view of

the world because that's where you have to
start we're still at that journey I think

that the perception on the outside even
among the Googles who have great advisers

they don't actually live in this world
they don't see all of the nooks and

crannies they have a simple construct for
how to transform it with data they

have incredibly sophisticated tools we
have incredibly sophisticated tools around

decoding D.N.A. We thought we
thought as soon as we had done

the whole human we that we are on our
way now that you know the race is on and

investment in the ninety's just
flooded in to genomic Samina

collapsed at the end of that decade it
was it was remarkable to watch right.

So that this is how
the cycles work that that.

Takes a while and so I think we're
we're we're probably in the second.

Phase I don't think we'll get there in
three My guess is we'll get there in

four or five but there's there's still a
lot of naivete if you and I went to Google

to the best people and we said What
are you working on it would probably take

three questions to expose that they have
no idea how they're going to connect a.

Really great presentation you.

Just have that's a quick question for
going to jobs to be done about

Clayton Christensen and
he came up with that he came up with it

because he felt that many
managers rely too much on.

Their financial statements and
emphasize automation and

I don't think that we've
noticed here is sometimes.

It can kind of dull the senses with
regards to doing hands on me to do so

in my severely constrained
just like it's done for

many people in finance if you're a public
company on the kinds of innovation you're

going to do because you're so reliant on
data any so it's a very good question and

it reveals the importance of so
it's important to read both your work and

question some because they have
different ways of talking about it.

Because she's in more
ephemeral more concrete but.

Both make the point that jobs
have both functional motional and

social and if you've invented
something that put somebody to sleep.

Then you you've left something really
important out of your invention and

the invention was probably forced on
you so the electronic health record

if we said to doctors you're going
to love this and please adopt it.

We would still be in two thousand and
it had to be forced on them even though

we all knew it was a completely inadequate
technology completely inadequate because

we couldn't figure out a way to get the
ball started towards digit digitization

the problem now is how do
I If you're a doctor or nurse

how to wire I develop applications
that allow you to work at the top your

license How do I take off and automate
things that you don't like doing and

that you shouldn't do and so when we think
about helping somebody get their job

done better we have to
really clarify like for

a doctor why are you here I shouldn't be
in the first instance asking you questions

you should come in and lots of data
should show up and I understand.

And get a gestalt of why here and then
I ask questions of top of that I guess

now I'm getting deeper into my
expertise to sort of understand so this

connection between jobs and automation
really is around the intelligent ways in

which we do it that allows people to
work at the top of their training and

then even optimize them from their very.

Concrete career advice because many of
us are through as the beginning of the.

Next become of it it's time to be in
a tech company like Google Facebook

or going to.

Biotech or house systems.

Now but you know so
much of of this business and

if you think OK I'm just
graduated this past year from.

Knowing the shooter and
I want to change of healthcare

what should they do next or
if this is a suggestion.

About a hundred.

And you know I could probably be
expounded on of overheard a drink or

two and I don't know.

I mean I think of.

I think of big things like patient
navigators and I think that this is a this

is an idea that really is transformative
that if we had navigators that.

To take on college care let's just
take one area let's take a woman

has a mammography and there's an
abnormality that requires follow Christ so

the journey her journey begins at
that point she is on that journey for

the first time so she has no idea.

What this journey is about right it would
be like you do in the old days we did this

with a map where we didn't we just
sort of drove but you need to get to

Lewisburg right how are you going
to get there if you have no idea if

that's what the journey is after after
an admiral and all the steps along

the way anticipating as just as the
application is talking to your record and

informing you about what to expect
an engine that would be gigantic and

that's transformative but initially
there's probably dozens of those kinds

of gaiters enough to get and
it's a huge data challenge and

I'm it's a huge communication problem
it's it involves translating a lot of

it's involves predicting it involves lots
of what's going on so there's lots of

there's lots of areas of our business
where I think these kinds of things

could be invented maybe it's a it's
a after it was really you know this

super helpful status figure when work or.