>> Mary Engler: Well, welcome back from break and

I'm delighted to introduce -- after such an incredible

morning with such great speakers -- I'm delighted to

introduce our next speaker Dr. Bonnie Westra, who'll be

presenting Big Data Analytics for Healthcare.

Dr. Westra is director for the Center of Nursing

Informatics and associate professor in the

School of Nursing at the University of Minnesota.

She works to improve the exchange and use of

electronic health data.

Her important work aims to help older adults remain in

their community and live healthy lives.

Dr. Westra is committed to using nursing and health

data to support improved and better patient outcomes as

well as developing the next generation of nurse

informaticists -- informatistatcians.

[laughter]

Okay.

Please, join me in a warm welcome for Dr. Westra.

[applause]

>> Bonnie Westra: Is it potato or potato [laughs]?

[laughter]

So, I am just absolutely thrilled to be here and this

is an amazing audience.

It's grown since last year, so this is great.

So, today what I'd like to do is to relate the

importance of big data in healthcare to what we're

talking about today, identify some of the

critical steps to make data useful so when you think of

electronic health record data or secondary use of

existing data, there is a lot that has to be done to

make it useable for purposes of research.

Look at some of the principles of big data

analytics and then talk about some examples of some

of the science, and you'll hear a lot more about that

during the week in terms of more in depth on that.

So, when we think about big data science, it's really

the application of mathematical algorithms to

large data sets to infer probabilities

for prediction.

That's the very simple definition.

You'll hear a number of other definitions as you go

through the week as well.

And the purpose is really to find novel patterns in data

to enable data driven decisions.

I think as we continue to progress with big data

science, we won't only find novel patterns but in fact

we'll be able to do much more of being able to

demonstrate hypothesis.

One of my students was at a big data conference that

Mayo University in Minnesota was putting on, and one of

the things that they're starting to do now is to

replicate clinical trials using big data, and they're

in some cases able to come up with results that are 95

percent similar to having done the clinical

trials themselves.

So we're going to be seeing a real shift in the use of

big data in the future.

So when I think about big data analytics, what this

picture's really portraying is big data analytics exists

on a continuum for clinical translational science from

T1 to T4 where there's foundational types of work

that need to be done but we actually need to apply the

results in clinical practice and to learn from clinical

practice that it then informs foundational

science again.

When you look at the middle of this picture, what this

is really showing is that this is really what nursing

is about.

If you look at the ANA's scope and standards of

practice on the social policy statements, nursing

is really about protecting, promoting health and then to

alleviate suffering.

So when we focus on -- when we think about big data

science in nursing, that's really kind of our area

of expertise.

And what you see on the bottom of this graph is it's

really about when we move from data, you know, we

don't lack data.

We lack information and knowledge and so it's really

about how we transform data into information into

knowledge, and then the wise use of that information

within practice itself.

This was, I -- we were doing a conference back in

Minnesota on big data and I happened to run into this

graphic that just, you know, it's like how fast is data

growing nowadays?

And so what you can see is data flows so fast that the

total accumulation in the past two years is a zeta byte.

And I'm like, "Well, what is a zeta byte?"

A zeta byte is a one with 21 zeroes after it.

And that what you can see is the amount of data that

we've accumulated in the last two years equals all

the total information in the last century.

So the rate of growth of data is getting to be huge.

Data by itself though, isn't sufficient.

It really needs to be able to be transferred or

transformed into information and knowledge.

Well, when we think about healthcare, what we can see

is that the definition is that it's a large volume,

but it might not be large volume.

So when you think about genomics sometimes it's not

a large volume, but it's very complex data, and that

as we think about getting beyond genomics and we think

about where we're at, it's really looking at where are

all the variety of data sources and, it's the

integration of multiple datasets that we're really

running into now.

And it's data that accumulates over time, so

it's ever changing and the speed of it is

ever changing.

What you can see in the right-hand corner here is

that there -- as we think about the new health

sciences and data sources, genomics is a really

critical piece, but the electronic health record,

patient portals, social media, the drug research

test results, all the monitoring and censoring

technology and more recently adding in geocoding.

So as we think about geocoding, it's really the

ability to pinpoint the latitude and longitude of

where patients exist.

It's a more precise way of looking at the geographical

setting in which patients exist, and that there's a

lot of secondary data then around geocodes that can

give us background information about

neighborhoods that include such things as, you know,

looking at financial class, education.

Now it doesn't mean that it always applies to me,

because I might be an odd person in a neighborhood,

but it gives us more background information that

we may not be able to get from other resources.

So, big data is really about volume, velocity, voracity

as Dr. Grady pointed out earlier today.

Now as we think about big data, 10 years ago when I

went to the University of Minnesota and my Dean,

Connie Delaney [phonetic sp] had talked about doing data

mining and I thought, "Oh, that sounds

really interesting."

Because I was in the software business before and

our whole goal was to collect data in a

standardized way that can be reused for purposes of

research and quality improvement.

I just didn't know what to do with it once I got it.

And so I've had the fortune to work with data miners.

We have a large computer science department that does

internationally known for its data mining, and a lot

of that work was funded primarily by the National

Science Foundation at that time because it was really

about methodologies.

Well now we're starting to see big data science being

funded much more mainstream in addition now, NIH, CTSA,

et cetera, are all working on how do we fund the

knowledge, the new methodologies that we need

in terms of big data science?

So, an example of some of the big data science that

really is funded already today is that if we look at

our CTSAs.

So, there's 61-plus CTSA clinical translational

science awards across the country and the goal is to

be able to share methodologies, to have

clinical data repositories and clinical data

warehouses, and then to begin to start to say, "How

do we do some research that goes across these CTSAs?

How do we collaborate together?"

Or as we look at PCORnet.

PCORnet is another example.

So as we think about, there are 11 clinical data

research networks -- this may have increased by now --

as well as 18 patient powered research networks.

We happen to participate in one that has 10 different

academic of healthcare systems working together,

and it means that for our data warehouse we have to

have a common data model with common data standards

with common data queries in order to be able to look at

research such as we're looking at ALS, obesity, and

breast cancer.

And wouldn't it be nice if we could look at some of the

signs and symptoms that nurses are interested in

addition to looking at specific kinds of diseases?

When we look at some of the work that Optum health as

well as other insurance companies, they're really

beginning to take a look at amassing large datasets.

So Optum Labs happens to have 140 million lives from

claims data, and they're adding in 40 million lives

from electronic health records, so that provides

really large data sets for us to be able to ask some

questions in ways that we haven't been able to do.

I'm excited about reuse of existing data, and so

hopefully some of that enthusiasm will wear off on

you today because it's really a great opportunity.

Now, in order to use large data sources, what that

means is that we need a common data model.

We need standardized coding of data and we need

standardized queries.

What I mean by that is that if we don't ask about the

same variables and we don't collect the data or code the

data in the same ways, it makes it hard for us to be

able to do comparisons then across software vendors or

health systems or academic institutions.

And with the PCORI grant for instance, we're actually

looking at how do we do common queries so that if

we've got the common models, we can write a query and

share the queries with others to be able to pull

data out from multiple health systems in a

similar way.

So I'm going to talk about what I mean by that a little

bit more and show you examples of how we have to

be thinking in nursing about this as well as thinking

interprofessionally.

So when you look at PCORnet, they started with a common

data model one, then they went to version two, and now

this is version three that's being worked on at this time.

So you can see in the top left hand corner we have

conditions which might be patient reported conditions

as well as healthcare provider conditions, but you

can also see that down in the left hand corner that

there are also diagnosis.

So diagnosis are ICD9 coding that goes with it.

ICD10 is that unfolding.

Notice when you think about your science, where is the

data that you want for your science, and is it

represented in this common data model?

I would suggest that there's many types of data in the

common data model that's important to all of us as we

think about where we're going whether it's

demographics of medications or, you know, what are the

kinds of diseases that people have?

And there's also something missing as we move forward.

So, before I get to what's missing one of the things

that I want to point out that's critical is that in

order for PCORI or NCATS or any of these other

organizations to be able to do queries across multiple

institutions they have to have data standards.

And so when we look at demographics for instance,

OMB is the standard that we use for demographics.

When we look at medications, it's RxNorm for medications.

Laboratory is coded with LOINC.