00:03
hello everyone
00:04
welcome to this training session on
00:07
advanced gis methods
00:09
and this time we're going to be looking
00:11
at our heart and multi-dimensional
00:13
indices
00:14
and this is part one so there's uh two
00:16
videos to watch this is first one
00:18
where we're looking at the aha data set
00:20
specifically so access to healthy
00:22
assets and hazards i'm dr nick verman
00:26
and i'm run training sessions for the
00:29
the cdrc
00:30
and i'm hosting today and i'm joined by
00:32
uh dr mark green
00:33
from the university of liverpool he's
00:35
going to say a bit about
00:37
our heart and tell us all about it and
00:39
mark and i are going to have a bit of
00:40
discussion
00:41
as we go through wonderful welcome mark
00:44
excellent thanks for having me today and
00:46
hello and welcome to everyone watching
00:47
the video
00:48
i hope you hope you're all staying safe
00:49
wherever you might be today
00:52
okay so
00:56
we'll start with um a little
00:58
presentation
00:59
so nick has kindly asked me to talk
01:02
today and introduce our access to
01:04
healthy assets and hazards
01:06
results otherwise we refer to as aha
01:10
um so what we're going to do is we're
01:13
just going to kind of go through what it
01:15
is
01:15
what it can be used for how it was
01:17
created and we'll talk a little bit
01:19
about some of the strengths and
01:20
weaknesses
01:21
of the resource and and there'll be a
01:23
bit of discussion between me and
01:25
nick over some of these things so let's
01:28
start with what do we mean
01:30
what is aha so a bit of background so
01:33
as part of the consumer data research
01:35
center which is an
01:36
esrc investment and really the
01:40
cdrc was kind of set up to try and open
01:42
up
01:43
data and opportunities to kind of
01:46
connect researchers
01:47
academic researchers to more kind of
01:49
private industry
01:51
consumer firms who have often quite
01:54
impressive or interesting data that
01:56
might have some value to
01:58
public society so this might be things
02:00
like
02:01
loyalty card data it might be more kind
02:03
of
02:04
official records and that we might be
02:07
able to use in different ways
02:09
and to answer some kind of society's
02:12
most pressing problems
02:13
so the cdrc was sort of set up as a
02:15
front to try and
02:18
kind of act as a kind of middle person
02:20
to broker some of these relationships
02:22
and provide data access to some of these
02:24
sensitive
02:25
and useful data sets
02:29
as part of that kind of cdrc project
02:32
what we've also been doing is kind of
02:33
processing some of these data that we
02:35
can't necessarily openly share and
02:37
trying to
02:38
to derive indicators from them or small
02:41
area measures and these are across a
02:43
whole range of topics
02:45
so for example we've been doing a lot of
02:47
work around trying to look at retail
02:49
dynamics looking at
02:50
high street sales looking at high street
02:53
location performances
02:55
there's a whole strand of research
02:57
looking at things around like how do we
02:59
tell individuals ethnicity from their
03:01
names which might be important for
03:04
example in hospitals where they don't
03:06
necessarily know them but they might
03:07
want to know
03:08
differences and inequalities and risk
03:10
and outcomes
03:13
and also there's been a whole suite of
03:15
indicators around kind of health healthy
03:17
environments
03:18
and which is what we're really going to
03:19
talk about today
03:21
so within that kind of health indicators
03:23
team we've been
03:24
developing what we call the aha project
03:27
which is a collaboration between myself
03:30
and professor alex singleton and alec
03:32
davies who are all based in the
03:34
geography department at the university
03:35
of liverpool
03:36
we also have um we also worked with
03:39
costas darris and benbahu in public
03:41
health
03:42
again at the university liverpool and
03:43
we've also had a lot of input from
03:45
public health england
03:47
around designing some of the kind of
03:48
health indicators work that we've been
03:50
doing
03:51
so why was we interested in kind of
03:53
looking at health and particular healthy
03:55
environments
03:58
well for example let's take this
04:00
individual
04:01
now we might be interested in what
04:03
drives this individual's health we might
04:05
talk about their diets their physical
04:06
activity
04:08
but they aren't you know they live in
04:11
the world around them and they don't
04:12
live in some sort of social vacuum they
04:14
interact with the world
04:16
that's around them and that might be
04:18
whether they live in an
04:19
area that has polluting industries that
04:22
might mean that
04:23
air quality is much poorer in their
04:25
environment which might then
04:26
impact on their respiratory health
04:29
it might be that access to health
04:32
services whether that's something more
04:34
formal like gps hospitals and of course
04:37
it might be that and if an individual is
04:39
really far away from their gp they might
04:41
put off going to see them as quickly as
04:43
possible
04:44
because it's just a bit more of a faff
04:46
to get there
04:47
but it could be whether they live in an
04:49
area has lots of gyms leisure services
04:51
that might
04:52
encourage them to be physically active
04:55
and of course it might be linked to the
04:57
retail environment around them it could
04:59
be
04:59
around whether they have healthy food
05:01
that's both accessible
05:03
and affordable around them it might be
05:05
if they live in an area of lots of pubs
05:08
that they're more or less likely to
05:10
drink as well
05:11
and it could be a whole range of things
05:12
around supermarkets casinos
05:16
these environmental influences might
05:18
have an impact on people's behaviors
05:22
but even if this kind of concept around
05:24
how the neighborhood might
05:25
shape and our accessibility to features
05:27
in the neighborhood might shape our
05:29
behaviors
05:30
it's also something that's really
05:31
important to policy makers as well
05:34
in particular modifying environments and
05:36
neighborhoods is something that's really
05:38
important and something that we can
05:40
feasibly do as local and national
05:42
governments
05:43
whether that might be local governments
05:45
restricting the location of new
05:47
fast-food outlets whether it's around
05:49
building more green spaces
05:51
or whether it's about trying to
05:52
understand the equitable accessibility
05:55
to health services for planning
05:57
new location of gps or pharmacies
06:02
so it seems like geographic context the
06:04
neighborhood really matters
06:06
at least might influence our health but
06:08
it also matters for a policy perspective
06:11
but at the same time it's quite tricky
06:13
to get hold of these data often these
06:15
data aren't exactly open
06:16
they might be stored and by data
06:19
providers who don't share it sometimes
06:21
they're stored by local governments who
06:22
just don't talk to each other so you
06:24
don't have a national
06:26
level data set and often these data come
06:28
in really awful formats they just need
06:30
processing
06:31
they need um kind of creating into a
06:34
much more user-friendly concept for them
06:36
to be able to use
06:38
so with those kind of issues in mind we
06:40
set out to try and create
06:42
um kind of the most comprehensive open
06:45
database of small area indicators around
06:47
healthy environments
06:49
health related features of environments
06:51
and making that all openly and freely
06:53
accessible to researchers to the public
06:56
and so they can have start to have a
06:57
look at how their environments are they
06:59
can start to see
07:00
whether there are links to people's
07:02
health health issues in their local
07:04
areas
07:05
and whether there's sort of actionable
07:07
objectives that they can use to push for
07:09
change in these environments
07:12
and then we also set out to try and
07:14
develop a summary index
07:16
and so a multi-dimensional index that
07:18
would summarize all of these features
07:20
and environments at the same time
07:23
and around trying to kind of
07:26
provide something that's a bit more easy
07:28
manageable way of targeting
07:30
maybe unhealthy environments
07:33
and what we ended up with is our kind of
07:35
access to healthy assets and hazards
07:37
resource
07:39
which is hosted on the cdrc platform so
07:42
you can find that on the maps
07:43
platform which we'll come back to a bit
07:45
later in this talk as well as the data
07:47
all being openly accessible on the cdrc
07:49
data server
07:53
so what can it be used for so we have
07:56
quite a range of data and there's a lot
07:58
of data opportunities because we've
08:00
created quite a suite of measures that
08:02
might be useful for different contexts
08:05
in particular if we're interested in
08:06
looking at accessibility
08:08
to certain stores of certain features of
08:11
environments
08:12
that might be a really good way of
08:13
trying to understand what are the
08:15
problems facing our populations
08:17
and so for example on this slide we're
08:19
just looking at
08:20
the median distance so the average
08:22
distance
08:24
of a area or a postcode within an area
08:27
to its nearest pharmacy pub gp
08:30
dentist and what we can see is actually
08:33
most of our population is quite is quite
08:35
accessible to many of these services
08:37
that might be good or bad for their
08:39
health
08:40
so most people are less than one
08:42
kilometer
08:43
to their nearest pharmacy and but about
08:47
44
08:48
of our population in great britain is
08:50
within one kilometer of a pub or a gp
08:53
which probably says
08:55
something wonderful about british
08:56
society
08:58
um but of course this might then give us
09:01
an
09:01
idea of trying to understand well which
09:03
types of people which populations are
09:05
more or less
09:07
have greater exposure to some of these
09:10
facilities so if we take for example
09:13
fast food outlets as one of our examples
09:16
and we can look at the distance to a
09:17
nearest fast food out there
09:19
and we can split it by level of
09:21
deprivation
09:22
using the index for multiple deprivation
09:25
in this case we're just looking at
09:26
decimals so
09:27
10 equal bands for areas where one is
09:30
the most deprived areas and 10 are the
09:32
least deprived areas
09:34
actually those areas in the least
09:35
deprived decile are twice as far away
09:38
from their nearest
09:39
fast food outlet and compared to the
09:40
most deprived areas which might
09:42
try to help us to understand health
09:44
inequalities in particular
09:46
a lack of healthy access to foods in the
09:49
most deprived areas
09:50
that might go one way to explaining
09:53
social inequalities in obesity
09:57
our data might be useful for linking to
09:59
social surveys linking them
10:01
to kind of other data sets of
10:04
administrative data sets and we've
10:06
looked at this ourselves
10:07
trying to link our kind of overall index
10:09
as well as some of the
10:10
other indicators from the original
10:12
version of aha
10:14
and what we've found is that sometimes
10:16
there's associations of health and
10:18
sometimes they vary in what they're
10:19
associated to
10:20
so for example the overall index we
10:22
don't find
10:23
necessarily any association to an
10:25
individual's physical health
10:28
but we do find that areas and people who
10:30
live in areas with much
10:31
poorer have overall scores also have
10:34
poorer well-being at the same time
10:37
and we can start to split that out by
10:39
all of these specific indicators so when
10:41
we look at the environment domain people
10:43
who live in
10:44
areas that have less green space and or
10:47
poor air quality they also have poorer
10:49
physical health at the same time
10:51
this might be important for them trying
10:53
to elucidate the drivers and
10:55
determinants of health
10:57
amongst individuals but also amongst
10:59
areas at the same time
11:03
from a more applied local government
11:05
aspect it's been used by quite a few
11:07
local governments around
11:09
targeting areas and trying to understand
11:10
the state of inequalities in their local
11:13
environments
11:14
so for example the um tower hamlets
11:17
local authority we're using aha our
11:20
overall index to just to make a case for
11:23
how unhealthy most of the areas in their
11:25
local area
11:26
are and in particular saying that
11:29
majority of their population really live
11:31
in areas that classify as quite
11:33
poor on our heart index compared to
11:35
other london boroughs at the same time
11:38
and therefore trying to make that case
11:40
that they need to intervene in
11:42
environments
11:42
to try and change those to make them a
11:45
bit more positive
11:47
of course that could be the over that's
11:50
the overall index but also more specific
11:52
individual indicators that we've created
11:55
have also been used by many local
11:57
governments for example gloucestershire
11:58
who were using it to monitor air quality
12:01
and map that out to try and see which
12:03
population which
12:04
area have poorer air qualities and
12:08
higher air pollutants to try and make
12:11
that case again for why they need to
12:13
try and tackle these issues so some of
12:16
this data can be used for both for
12:17
monitoring
12:18
aspect as well as trying to understand
12:20
some of the issues facing their local
12:22
populations
12:25
so thanks mark that's uh it's great to
12:28
see
12:28
some examples of what what the index can
12:31
be used for
12:32
particularly in local authorities and it
12:34
it sounds though you're trying to
12:36
collect a huge amount of different data
12:37
together and to create
12:39
a relatively easy to use index
12:42
is that it's the easy to use aspect
12:45
something that the local authorities
12:46
have liked so it's allowed them to
12:48
actually
12:49
get grapple on this data
12:52
yeah i mean i think one of the biggest
12:55
issues we faced when we started talking
12:57
to local authorities about what they
12:59
might want
13:00
was that they don't necessarily always
13:02
have the technical and skills capacity
13:05
to generate these indicators themselves
13:07
so
13:08
for example um when we met with
13:10
liverpool city council they had all of
13:11
the data on the location of fast food
13:13
outlets they had the location
13:15
of gps they had a lot of that kind of
13:19
air quality
13:20
monitoring information but they didn't
13:22
necessarily have the kind of skills to
13:24
process it
13:25
for generating things like well how does
13:28
access to these features vary across
13:30
their local areas
13:31
so what they really benefited from was
13:33
actually was saying well we've got this
13:35
technical
13:36
setup that we could just generate these
13:37
data for what you need
13:40
um and making that openly available
13:43
allowed them
13:44
to kind of just use that there was no
13:46
barriers to it was very quick and easy
13:48
and of course we set out with you know
13:50
appropriate metadata to try and make it
13:52
clear what what is available as well
13:55
so was was there kind of input
13:58
quite important in terms of how you
14:00
constructed the index
14:02
as well in terms of what was important
14:04
and what wasn't
14:06
yeah so i mean we had public health
14:07
england as an advisory board but we also
14:09
talked to a few local authorities about
14:11
what they'd like to see in that
14:13
the set of indicators that we we've
14:15
created
14:16
and so we had a bit of input and you
14:18
know i think the biggest thing they said
14:20
was well
14:21
we just like to see a lot of these
14:22
indicators because we we simply don't
14:24
have the data
14:25
often and it's something we've always
14:28
been quite open about with our kind of
14:30
aha resources actually that
14:32
if there's something missing or people
14:33
want something creative that we probably
14:35
have the bandwidth to create it we've
14:37
got a lot of that technical
14:38
setup now so if people get in contact
14:40
and say hey we're
14:41
really interested in this issue well if
14:44
we can create it we will because
14:46
you know we like to think that we're a
14:48
good resource that um
14:50
at least matches what people think are
14:52
useful rather than
14:54
you know ourselves and that was also you
14:57
know that process around what was needed
14:59
was also supported by a kind of general
15:01
scoping review across the literature to
15:03
see what were
15:04
important aspects of environments to
15:06
measure so going from that evidence as
15:08
well as talking to stakeholder groups
15:12
yeah and if you're listening to this and
15:15
uh do want more information or do have
15:17
suggestions about improvements there
15:18
will be some contact details
15:20
below the video so please do get in
15:22
touch and
15:23
let us know what would be useful for you
15:26
as well
15:27
i think now might be a great time to go
15:29
into kind of how you constructed
15:31
the index a bit more because so far
15:34
we've got
15:34
uh there's a there's a huge amount of
15:36
data which you alluded to at the
15:37
beginning
15:38
but how do you kind of combine that into
15:41
the index that we've talked about
15:43
awesome okay so we'll go on to now
15:47
how we created what we did and what we
15:49
do have available
15:52
so what we kind of came up with is this
15:56
general framework
15:57
in that what was interesting is how do
16:00
we create an indicator and how do we do
16:01
that in
16:03
you know a fairly robust way
16:06
so for example we may let's take the
16:08
example that we
16:09
want to look at the travel time to your
16:12
nearest gp
16:13
for a local area so we identify that
16:16
particular indication that comes out of
16:18
the
16:19
process of talking to stakeholders as
16:21
well as that kind of literature review
16:23
process
16:25
what we do is then we try to access that
16:27
data and in case the gp
16:29
records we get them from nhs digital for
16:32
england wales
16:33
and then the national records of
16:35
scotland have their gp locations
16:38
for scotland and so we collect
16:41
information that has the location of all
16:44
gp
16:44
and surgeries their location
16:48
and then we also get hold of all
16:50
postcodes for
16:51
every area in great britain so
16:55
um talking about about 1.75 million
16:58
postcodes
16:59
and using those kind of two sets of
17:03
information so we have postcodes of
17:04
individuals and areas
17:06
we have postcodes of the kind of service
17:08
and gp
17:10
what we do is we do a very simple step
17:12
first we take for every postcode we draw
17:14
a little buffer around it
17:16
and then we say right we're just going
17:19
to focus on the nearest 10
17:21
gps within that buffer so we
17:24
we define an arbitrary distance and we
17:27
define an
17:28
arbitrary number of gps to look at and
17:30
but we do this first step
17:32
really because it saves on a lot of
17:34
computational next step
17:37
the next step then is once we've got our
17:39
kind of ten nearest
17:40
roughly nearest um gps is we use
17:44
something called routinu so routino is
17:47
an open source
17:48
piece of software that allows us to
17:49
calculate travel time from two points
17:53
and we supplied routino with the kind of
17:57
road network distances or public
17:59
pavements
18:00
as well from open street map which
18:03
allows routinely then to calculate using
18:05
a lot of information around
18:06
speeds and mode of transport in our case
18:10
we just use
18:11
car travel time and to work out
18:14
the time and distance to a nearest set
18:17
of other points in our case
18:18
from postcodes of individuals to um
18:22
postcodes of gps the reason we took
18:25
those kind of first step of
18:26
subsetting um the newest 10 gps is
18:31
because we could have done the nearest
18:32
travel time to all
18:34
gps but that would just take forever and
18:37
when whilst it's only 1.75 million
18:39
postcodes
18:40
it's taking really about eight hours
18:43
processing time
18:45
and to do the whole country and which is
18:47
quite a significant effort
18:49
really um even on our kind of more bps
18:52
servers
18:55
so what we do is we we calculate then we
18:57
we identify the nearest
19:00
gp and then we store that information so
19:02
for that each postcode of
19:05
um in great britain we have the travel
19:07
time and distance to its nearest general
19:10
practice
19:11
and then we take that information and we
19:14
aggregate those up to various different
19:16
geographical zones in our case we use
19:18
lower super output areas for england
19:20
whales and data zones for scotland so
19:22
they're roughly equivalent
19:24
all those data zones are slightly
19:26
smaller
19:27
and essentially we just take a median
19:28
value we take the the average
19:31
which then gives us a kind of summary at
19:34
that
19:36
that spatial scale and that was kind of
19:38
important because we could we can
19:40
i mean we could make the postcode
19:41
information available it is a controlled
19:43
data source by the cdrc
19:46
but it's not necessarily particularly
19:47
useful whereas partic
19:49
showing it at kind of a neighborhood
19:50
level it's a bit easier for people to
19:52
grasp and it's also the tends to be the
19:55
spatial scales that
19:56
local authorities are using so it's kind
19:58
of designed with our
19:59
stakeholders in mind
20:02
now now's probably a good time to
20:03
mention that when if you're new to using
20:06
gis data you will come across terms like
20:09
lsoa
20:10
and data zone and there's a whole suite
20:12
of different things
20:14
so if these are new to you do look them
20:16
up and there's various resources
20:17
available that explain
20:19
what they all are so was the the
20:21
selection of lsoas that was informed by
20:24
the
20:24
the local authorities that you worked
20:26
with to a degree
20:27
was that right mark yeah yeah is
20:29
basically that's what they said was the
20:32
most common statistical sort of
20:34
neighborhood zones that they were using
20:36
so it was to try and match what they
20:38
thought was useful
20:40
okay great great thank you okay
20:44
so we'll get back so what we did is we
20:47
did that sort of process for a series of
20:49
indicators
20:50
um and we kind of grouped them into
20:54
some general domains we'll come to the
20:56
domains in a minute
20:57
and so what we did is we calculated
20:59
distance to nearest fast food outlet
21:01
pub bar or nightclub gambling outlet
21:04
tobacconist off license and which we all
21:08
kind of
21:09
group them together as part of the
21:10
retail retail services essentially
21:14
we then looked at a series of health
21:15
services including gp surgeries
21:17
hospitals with a
21:18
e services dentists pharmacies we stuck
21:22
leisure centre and gym with those in
21:24
this kind of grouping of domains
21:25
simply because it was more of a positive
21:28
part of the environment they didn't seem
21:30
to fit
21:30
you know in the retail services there
21:32
were more negative aspects of health
21:36
we also um had an air quality series of
21:40
measures now with this we didn't
21:41
calculate distance and accessibility
21:43
because it really make any sense
21:45
rather we took some of the deferral
21:47
estimates on air pollutants and we just
21:49
calculated mean values based upon those
21:52
estimates for the same areas
21:54
we ended up picking three pollutants
21:57
sulfur dioxide particulate matter pm10
22:00
and nitrogen dioxide and because they
22:04
ended up being what we thought from the
22:05
literature were most important
22:07
and a lot of these pollutants are kind
22:09
of highly correlated so
22:11
we couldn't stick every one of them in
22:13
because it would cause a lot of issues
22:14
around multiculinarity
22:16
further down the line we then have a
22:19
final domain which we call the natural
22:21
environment which has a kind of mixture
22:22
of mean
22:23
and distance values we have mean value
22:26
of green space within an 800-meter
22:28
buffer of a postcode
22:29
which we just any amount of green space
22:31
which we kind of call the passive
22:33
passive role it's the green space
22:35
whereas you're moving through an area
22:37
that you might benefit from um
22:40
and then we have a series of distance to
22:42
nearest measures including distance to
22:44
nearest green space
22:45
but being a bit more narrow active
22:47
places that you might be going to parks
22:49
grasslands might be going for a run or
22:51
to meet
22:51
friends actually it's really about
22:53
engaging with it rather than say walking
22:55
through a leafy
22:56
suburb and also blue space lakes rivers
23:00
and all the sea
23:03
so we we generate quite a lot of these
23:05
measures again based upon what
23:06
stakeholders suggested
23:08
and and also from the literature
23:11
so we have a series of indicators and we
23:13
group them into these kind of four
23:15
domains
23:16
and this is where we start to get
23:17
involved in the kind of
23:18
multi-dimensional index creation which
23:21
will be covered
23:22
in the second part of the video in a bit
23:24
more detail
23:25
so what we do is we take all of the
23:26
indicators for example in the retail
23:28
domain
23:29
we standardize them across the whole so
23:31
they're all kind of on the same
23:33
plate level level playing field and then
23:36
we take a kind of average value
23:37
essentially which we call our retail
23:39
average domain score we do that for each
23:43
of the four domains
23:44
and then again we aggregate them up so
23:46
we kind of standardize them
23:48
and then take an average score which
23:49
becomes our overall index
23:51
uh which we call the aha index and
23:54
that's how we get from kind of producing
23:56
these
23:56
wonderful diverse indicators towards
23:58
getting an overall
24:00
measure of how um kind of what the
24:02
environment is like across these kind of
24:04
four concepts
24:08
and and all of that data we've come
24:10
we've made it all openly available
24:11
so the code is available somewhere on
24:13
github for reproducing it
24:15
but we also have an interactive mapping
24:16
resource which allows you to engage with
24:18
that data directly
24:20
on the cdrc website and also you can
24:22
download the raw data
24:24
and for all versions of aha through the
24:26
cdrc website as well
24:28
and just to give you a kind of a flavor
24:31
of what that looks like nick has asked
24:33
me to to
24:34
give a kind of um a trial of the new
24:37
maps
24:38
uh website which isn't currently live as
24:40
we record this video so you're getting
24:41
something new and exciting in africa
24:45
um but just to kind of give you an
24:49
idea of what um it looks like
24:54
so so this is the um the cdrc
24:57
maps website which will be the new
24:59
incarnation the old one
25:01
it looks pretty similar but um has less
25:04
functionality
25:06
and but it allows you to kind of get a
25:08
qualm of a kind of general sense to
25:09
interact with that
25:10
that resource in one go and we've got
25:13
the aha data a lot
25:15
loaded in at the minute um and we can
25:18
just much like google maps zoom in zoom
25:20
out to areas
25:23
and it sometimes takes a minute to kind
25:25
of load in
25:26
aspects because it's got a lot of
25:28
information out of data kind of being
25:29
processed
25:30
we can have a look at the boundaries for
25:32
regions government regions we can add in
25:35
local authorities we can add in even the
25:37
buildings if we want as well
25:39
and it gives us a bit more of a way of
25:42
interacting with this
25:43
this data set and and again you can just
25:47
download all these data and create these
25:48
maps yourself it's all there to be
25:50
um to you be used but what we found is
25:53
this really works
25:54
kind of getting the public involved and
25:56
actually i really like this more
25:57
interactive mapping
25:58
aspect um because it makes it a bit more
26:01
real
26:02
and certainly those more public users
26:04
who've
26:05
emailed me about the excel and csv files
26:08
we provide for the raw data they really
26:10
struggle with that because it's
26:11
not necessarily set up quite as uh
26:14
helpfully
26:16
so in this case what we're doing is
26:17
we're looking at liverpool which
26:19
is where i'm based and we're looking at
26:22
our overall
26:22
aha index and it has on the right hand
26:26
side it has the kind of
26:27
legend which tells you in this case
26:29
we're looking at deciles where the blue
26:30
values
26:31
are and doing very well at least healthy
26:34
environments and the red are worse
26:35
performing so these are the poorer
26:38
environments
26:39
and i think what really struck us when
26:41
we created this initially because we
26:42
wasn't sure
26:44
um you always kind of when you create
26:46
something you're not sure whether it
26:47
will
26:47
tell you anything of value and but at
26:50
least seems to pick up things that we
26:51
might expect make sense in liverpool you
26:54
see a lot of red clustered around the
26:55
city center area
26:57
and that's this you know there's a lot
26:59
of kind of retail opportunities there
27:01
there's generally very few gpa
27:03
services there's no green space well
27:06
there's very little green space
27:08
so that's why that's going to come up as
27:10
bad but also it
27:12
picks up really subtle and inequalities
27:14
by space as well
27:16
so you can see that red kind of
27:17
continues north of the city centre
27:20
and also out east and these are more
27:22
deprived areas that tend to have
27:24
greater concentration of some of the
27:27
unhealthy retail aspects
27:29
and as well as kind of poor access to
27:31
health services sometimes and also poor
27:33
access to green space
27:35
but you can see it also picks up nice
27:37
parts in kind of south liverpool around
27:39
sefton park around chilled wall
27:41
so it's still picking up a kind of um
27:44
it's not just treating cities as the
27:45
same it's picking up quite a lot of
27:46
variants in the types of environments
27:48
and neighborhoods
27:50
people are exposed to
27:54
um is there anything you'd like to add
27:58
to that nick
27:59
yeah it's really fascinating to see and
28:02
it's it's great
28:03
um you know and i'd really recommend
28:05
that everybody do this if
28:07
if you are based in the uk there's an
28:09
area you know have a look at the map for
28:10
where you're based
28:11
and does that match your expectations
28:14
because you you have
28:15
often kind of knowledge about your local
28:16
area where's the more deprived areas and
28:19
less deprived and so on
28:20
um one thing i think it might be worth
28:22
adding is we can dive into the
28:24
individual domains
28:25
as well so the four different sections
28:28
and we can do that on the the maps site
28:30
i think um we can we saw the
28:33
the scores listed but you can pull up
28:35
the maps for each domain as well
28:37
um in terms of how it's actually used do
28:40
the
28:41
the people who've used the data so far
28:43
are they generally using the overall
28:45
index
28:45
or one of the four domains for the the
28:47
kind of work that they're doing
28:50
we kind of got mixed so we've got quite
28:52
a few people like that overall indicator
28:55
and because it's it's something that's
28:57
quotable or is something that's a bit
28:59
more
28:59
simple to present and what we find
29:02
though is also people
29:03
like to just focus on a specific
29:05
indicator so they'll just look at the
29:07
accessibility to fast food outlets for
29:09
example
29:10
because that's their specific focus
29:12
they're not interested in the other
29:13
aspects of this
29:15
and i think that flexibility and what
29:17
people can use it for is actually quite
29:18
useful
29:19
and and it's not just about here's one
29:23
indicator of how we think the world
29:24
works go and use it actually
29:27
because we make it all openly available
29:29
the inputs the outputs
29:31
we want people to use in however way
29:33
fits with what they need and you know
29:35
it's also
29:36
since the code's there it's more open
29:38
for people to
29:39
take it take it apart rebuild it and say
29:41
actually we think it's slightly better
29:43
at looking like this
29:45
and that competition is always good good
29:46
to have to be honest
29:48
yeah it's great to have that flexibility
29:51
so we have
29:52
the kind of top level index overall and
29:54
then we have the four
29:55
domains and then we have the individual
29:58
indicators as well
29:59
and are they all on the map site they're
30:02
all available
30:03
to download and they're all on the map
30:05
site as well so you can have a look at
30:07
fast food outlet accessibility you can
30:09
look at the pollution
30:11
all of that is on the map's um website
30:13
as well and
30:14
and that's where we've got a lot of
30:15
leverage from from the public
30:17
in particular who really like to just
30:19
click through and look at all of these
30:20
different aspects
30:22
and and they found it really interesting
30:25
actually
30:26
not just looking at that overall index
30:28
but actually just looking at specific
30:29
features of their environments and
30:31
trying to see
30:32
what are the local issues in my in my
30:34
area
30:35
and that strikes me it's a really
30:37
powerful tool
30:38
the the index itself but also the
30:40
visualization
30:42
so it's it's not just having the data
30:44
itself but it's being able to show
30:46
the stakeholders or whoever might be
30:48
interested this is a specific
30:50
uh indicator and these are some of the
30:52
issues we we need to
30:54
deal with at a local level or we need to
30:55
address yeah it makes the data more real
30:58
and i think the stereotype of a
31:00
geographer is
31:01
like we like to map everything but
31:03
there's a reason for that and that's
31:04
because
31:05
it's such an easy and intuitive way of
31:07
visualizing and sensing the world
31:10
and that kind of the classic the picture
31:13
can tell
31:13
a thousand words speaks volumes and maps
31:16
i
31:17
oh yes definitely yeah no i i definitely
31:20
agree with that i'm biased because i'm
31:21
also a geographer
31:22
uh but i i'd really emphasize that and
31:25
the
31:26
the power of showing data on a map is
31:28
really really useful
31:30
and there's lots of resources out there
31:32
if you've got
31:33
your own data there or stuff you want to
31:35
show on that yourself so please do check
31:37
that out as well
31:39
that's no that's that's great that's a
31:41
wonderful demo and uh great to see the
31:43
the new maps site in action
31:45
um i think the kind of last chunk it
31:47
would be good to look at now is what
31:49
what are kind of some of the strengths
31:50
and the weaknesses
31:51
of the the our heart index um because
31:54
we've heard
31:55
quite a lot of the uses um but there's
31:57
always
31:58
you know limitations with this so could
32:00
you tell us a little bit about those two
32:01
please
32:02
yeah sure okay so the final thing we'll
32:05
talk about is some of the strengths and
32:06
weaknesses
32:07
of this resource so i think
32:10
as a general strength it i think the
32:13
commitment to open source is
32:14
very important because it removes a lot
32:16
of those barriers
32:18
and it makes that data openly available
32:20
it's there for people to use and however
32:23
they might want to use it and again we
32:24
use
32:25
the inputs it's the indicators it's the
32:27
domain scores everything is openly
32:29
available and people can use what they
32:30
need they're not just
32:31
limited to one thing we have the
32:34
wonderful mapping resource which i think
32:36
is brilliant it just makes that
32:37
so much easier to engage with and you
32:40
can have a quick look at
32:42
kind of issues in your local area and
32:43
then download the data from more deeper
32:45
analysis
32:46
if you wanted to but it doesn't
32:48
associate any kind of technical
32:50
expertise
32:51
um and we don't think that you know we
32:54
have such a range of indicators on there
32:56
now that there's very few of the tools
32:58
that have such a
32:59
range of free open source um kind of
33:02
measures of healthy environments so we
33:04
think it's probably the most
33:05
comprehensive in great britain which is
33:07
is good and of course as i alluded to
33:10
before if there's stuff that you think
33:11
is missing from there where we've got
33:12
all of these methods
33:13
the infrastructure the technical
33:15
abilities it's all set up now
33:17
so we'd love to hear from you of
33:18
anything you'd like to see included and
33:20
you know
33:21
i think it's important that you know
33:23
it's not just what we think should go in
33:24
there it's more of an interactive
33:26
process
33:29
now i think the weaknesses really fall
33:31
into kind of two main areas and one of
33:33
them is around kind of technical issues
33:34
which i'll come to in a moment
33:36
the other one is around framing of this
33:39
and one of those kind of framing issues
33:40
is well
33:41
what do you really mean when you create
33:42
an overall indicator and what does that
33:44
actually mean
33:45
so if you've gotten a high score of 20
33:48
versus 23
33:50
what does that really mean and and
33:52
that's a it's a problem because
33:54
the actual index doesn't really have any
33:56
necessary meaning it has
33:58
the kind of variation matters so having
34:00
a high score versus a low score then
34:02
there's some value in that
34:04
because you're saying well this area is
34:06
better or worse than that area
34:08
but those kind of specific around how do
34:11
you interpret any of the numbers or the
34:13
ranks
34:14
and can be problematic
34:18
um there's a big difference between kind
34:20
of urban and rural areas and often this
34:22
is something we get a lot of discussion
34:24
around in that
34:25
our index sort of shows the rural areas
34:27
to be not always as healthy as we might
34:29
assume them to be
34:31
and often people say well they're really
34:33
green they're really
34:34
low air quality so surely they should be
34:36
quite healthy but
34:38
because obviously on our index they're
34:39
also located far away from
34:41
health services and often a lot of that
34:44
green space is just not simply
34:46
accessible and there's one thing about
34:48
living next to a
34:49
you know it's really nice to live next
34:50
to a farmer's field but you can't
34:52
necessarily go and engage with it you
34:54
can probably look at it and benefit
34:55
passively but not necessarily actively
34:58
um
34:59
but i think you know it's really about
35:00
opening up these discussions about what
35:02
do we value
35:04
um for health and environments really
35:08
a lot of discussion we get and this is
35:10
from doing a lot of the outreach of a
35:12
work with aha is that it's just not imd
35:14
it doesn't look exactly the same as the
35:16
index of multiple deprivation
35:18
which i kind of thought i put ground
35:20
there is because
35:21
we didn't set out to create the index
35:23
multiple deprivation because it already
35:25
exists
35:25
we set out to create something different
35:28
and
35:29
and so you know we shouldn't always
35:31
assume that these things are all
35:32
in linearly related um
35:35
but you know when you've got one measure
35:37
that kind of dominates things people
35:38
just
35:39
they like one index i think
35:42
once we so there's a lot to be said for
35:46
you know there there being one index uh
35:48
to measure
35:50
uh deprivation but we we need to
35:52
remember that
35:53
it's often kind of multifaceted so
35:55
there's lots of things
35:57
going on and you know the index of
36:00
multiple deprivation is a great resource
36:02
and it gets a lot of coverage
36:04
um but it's only measuring certain
36:06
aspects
36:08
and you know it's
36:11
trying to get that balance between how
36:13
much detail do we go into and how
36:15
easy to use do we make it and it is
36:17
often a kind of
36:19
uh tension between the two because we
36:21
can't
36:22
you know the world is not simple there
36:24
are these complexities in it there are
36:26
these kind of variations
36:28
uh that we have to take into account um
36:31
but it's trying to get that balance so
36:33
you know you probably use the imd
36:35
already
36:36
but this resource is different but you
36:39
can use them together
36:42
yeah and actually it's really
36:43
interesting when you look at the
36:45
interaction between the two measures and
36:46
you're looking at deprived areas that
36:48
have good and bad environments and
36:50
seeing
36:50
well does that influence differences in
36:53
health
36:56
um we have quite a large range of
36:58
measures but of course there might be
36:59
stuff that's missing
37:01
and from there as we discussed and we've
37:03
tried to be as comprehensive as possible
37:05
but the stuff that we kind of
37:06
miss out i think and of course it is
37:09
it's a descriptive tool it's just an
37:11
indicator
37:12
that measuring environments it's not
37:14
saying these environments are definitely
37:15
causing
37:16
bad health it's more saying well how do
37:18
we measure these environments to
37:19
actually look at how important they are
37:21
to to health really
37:26
some of the technical problems of it
37:28
well we i think
37:30
the biggest issue really comes down the
37:31
selection of weightings and how you
37:33
weight
37:34
the indicators and the domains now we
37:36
went about this by just rating them
37:38
equally so we treat the retail domain
37:40
the health domain the air quality
37:42
area air quality and the green space
37:44
domains it's just equal they are
37:46
the same we say they have the overall
37:48
same influence on overall indicator
37:51
but we've had quite a lot of criticism
37:52
about that saying well the retail domain
37:54
doesn't necessarily
37:55
influence health as much as the health
37:57
domain and
37:59
some things might be more or less
38:00
important the issue with that is that
38:03
there is no
38:04
evidence or nobody knows which is most
38:07
important and how we should weight them
38:08
and so
38:09
any waiting is kind of going to be
38:11
biased and unfair
38:13
and so we kept it as even
38:16
just simply because that was you know
38:18
the the fairest way we didn't have any
38:21
reasoning to to change the weightings
38:23
now
38:24
like i say we make all of the data
38:26
openly available so we're all
38:27
for people re-weighting it however they
38:29
want to and using it in the ways that
38:30
they want to
38:32
um to do yeah that that that's a great
38:35
point and we'll pick some of that up in
38:37
the practical sessions
38:39
so when you get to that point in the
38:40
practical you have the option of
38:43
rewaiting and we'll say a little bit
38:44
more about that in the next video as
38:46
well
38:48
and there's some general issues and that
38:50
maybe we don't provide
38:51
uncertainty of estimates and we're just
38:53
providing a point estimate maybe people
38:55
want error bars
38:57
on these things and we only have two
38:59
time periods at the minute 2016 2017
39:01
we're about to update to the most recent
39:04
um period of data we have one of the
39:06
issues is we can't go backwards we can
39:07
only go forwards which is a bit
39:09
difficult because if you want to link it
39:11
to historical data which is where it's
39:13
really interesting if you've got 10
39:14
years worth of data
39:15
that would be fascinating we just can't
39:17
do it at the minute we don't have the
39:19
data going further back which is
39:21
there's a problem one question we've
39:24
always wondered about the timeframes can
39:25
you do
39:26
comparisons so can you look at the data
39:28
in 2016 and in the same area in 2017 and
39:31
say
39:31
it's changed to get better or worse you
39:34
can
39:35
we did make some revisions to the
39:37
overall index so
39:38
we've increased the number of indicators
39:41
from 2016 data 2017
39:43
so you'd have to be careful in what
39:45
might be a change because of the
39:47
environment changing and what might be
39:48
because of
39:49
the changes in that methodology so we do
39:52
offer some caution to those kind of
39:54
comparisons
39:55
however um we are you know looking we're
39:58
not looking to make any revisions going
40:00
forward so in theory should be a bit
40:02
more stable
40:03
and some of those changes came out
40:05
discussions with public health england
40:07
after
40:07
the first release of our indicator
40:12
and i think the last comment i'll make
40:14
is really you know
40:15
all we're looking at is accessibility to
40:18
certain services
40:20
but it might not necessarily always be
40:21
the best way of looking at environments
40:23
it might be that we want to
40:25
look at the density of places it's not
40:27
just that you're
40:28
less than one kilometer away from fast
40:30
food out there it's that
40:32
you live in an area with 20 of them very
40:35
close proximity and actually that might
40:37
matter more
40:38
so how we try and measure environments
40:40
might
40:41
influence you know how we view the world
40:44
and how
40:44
we kind of get handled around these
40:46
health inequalities and we're only
40:47
really offering
40:48
one aspect of that we just kept it
40:50
simple and consistent
40:56
okay so that's really the end of of what
40:59
i was going to present on today and like
41:02
i say all of that data
41:03
is available openly and freely via the
41:06
maps website and you can be downloaded
41:08
from there as well
41:10
and we hope that you know it's a really
41:12
useful tool that can give a kind of very
41:14
detailed sense of small area issues
41:16
around healthy environments
41:18
and if you want any more further reading
41:20
around the kind of tall aha and the
41:22
resource
41:23
and there's three really nice papers
41:25
come out and
41:26
so we have one in health and place which
41:28
is open access
41:30
and which is really a technical details
41:32
around creating this multi-dimensional
41:34
index
41:35
so it goes through a lot of those more
41:36
technical details but also has that it
41:39
links to the social survey data
41:41
as well to see how useful it might be we
41:44
have a paper in scientific data
41:46
which really kind of trusts the the
41:48
whole of the resource
41:49
and goes through what we've created some
41:52
of it about how we created some of these
41:53
individual indicators as well
41:56
and we also have a kind of free um short
41:59
technical report that kind of goes over
42:01
data sources how they're created
42:03
which is a good kind of um thing to
42:05
refer to whenever you're using
42:07
the resource so thank you very much for
42:10
listening
42:10
and i'll hand it back to um nick
42:15
great wonderful thank you very much mark
42:17
that was uh you know
42:18
it's great to get a good overview of the
42:20
uh har index and
42:22
uh a few of the details of how people
42:25
are using it
42:26
uh but also the different components and
42:28
what we might be able to do with it in
42:30
the future
42:31
so there's so much potential there and i
42:34
really would uh recommend everyone to
42:36
certainly have a look at the maps
42:37
website have an explore and try
42:39
downloading the data
42:40
uh yourself and in the the practical
42:43
session that follows this we'll be
42:45
doing that so we'll be doing that in our
42:47
studio
42:48
um and creating some maps and giving you
42:50
the tools to actually
42:52
play about and explore so so please do
42:55
do that and do let us know how you get
42:57
on there's options to leave comments so
42:59
please do
43:00
and if you've got any thoughts or
43:01
questions please do drop the line
43:03
and we will do our best to answer them
43:08
wonderful thank you very much everyone
43:10
and i hope you enjoyed the
43:12
the session um i would recommend that
43:16
you also have a look at the
43:17
the second video in due course uh we'll
43:19
delve a bit more into the
43:21
method there so how we actually create
43:23
the multi-dimensional
43:25
um and how we can make some of those
43:27
changes so we'll talk a bit more about
43:29
the weightings we'll talk a bit more
43:30
about
43:31
how we combine the different domains and
43:33
why we fix that and data
43:34
and so on so there's lots more uh
43:37
interesting stuff in that
43:38
so we will see you in the next video
43:41
take care
43:47
goodbye