Today we will be liveblogging the Open Source, Geo and Health event taking place in Edinburgh. The tag/hashtag for the day is #gecohealth so if you’d like to join in discussions online please tag your tweets, blog posts etc. There is also a low res livefeed of audio here.
Because this post will be updated live there will be some spelling, typing and probably acronym errors – if you spot anything that should be corrected do let us know as we’ll be tidying up the post in the next few days and adding images etc. This should then act as a record of the day and a place to continue discussions in the comments.
The programme/headings for todays events is:
Welcome: James Reid, JISC GECO Project
JISC has funded a dozen projects around geospatial areas and GECO is there to help find the connections between those projects and to look at how geo connects into the much wider community within and beyond the academic sector.
The intersection between geo and health seemed quite timely and relevant.
You’ll have lots of time for discussion and participation today.
“Participatory Health Surveys” – Sergiusz Pawlowicz et al, Centre for Geospatial Science, University of Nottingham
I am based at the OGC CGS at University of Nottingham and was created in 2005 to really deal with anything related to geo and location. We work across the university. We have staff, researchers, students, and intern students there.
I think what we do is really web 2.5. Web 3.0 – something like Conrad Wolfram described – something generated by machines not people. So in our surveys we mix information from people with information from machines. We analyse their answers at the beginning and we try to deliver the realtime survey, different to each participant, depending on many streams of information we receive from various sources (geo located tweets, RSS etc).
Almost everyone has a smartphone which is GPS enabled so we can have good GPS location of participants. And internet enabled on the phone – but it’s not everywhere. And we need interoperability. And we decided to use HTML, not to deploy our own application for a particular handset, but to use HTML 5 which works on almost all modern smartphone browsers.
We add another dimension to these surveys. We have a survey knowledge survey that looks at the answers and asks more detailed questions based on those answers, or those of others in the community who have answered the survey.
There is automated noise reduction (to filter questions) and human sensors if you will – they provide information for us.
Twitter will launch next month basic system to filter tweets and that should help with deploying these kinds of tools in the future.
I decided to build this system using the Drupal system (open source), use standards (W3C, OGC), decided to use standard fonts etc. to ensure the tool was easy to use on phones.
The results can be displayed in real time and can be analysed later. I am rather concentrated on background and technology so the interface is rather simple.
The tools used let this be deployed quickly and privacy is a core factor here – we monitor privacy constantly to ensure that we keep data secure.
Q1 – Ananad) About the real time knowledge reduction. There is a risk in that approach is that you bias your incoming data. The statistical analysis of that can be very complicated. So for a survey of diahrea and vomiting that might be fine and not too much of a problem, but in other contexts, such as an outbreak of legionella, that have very
The knowledge base is separate from the main system for the surveys so you can choose whether to build a static survey or a dynamic survey and you can choose how the knowledge base interacts with the survey. So you could also approach this by running 2 surveys – an initial static survey and after more detailed analysis, deploy a second survey. In any case this will be better than paper and pencil
Q2) You have open data on one side and confidential data on the other. There may be valuable research data in the confidential data. How do you see that open/closed tension in a real world setting
A2) We store the sensitive data on a separate machine so the sensitive and other data do not mix. They are combined in the knowledge base machine but only to serve dynamic surveys. We can logically separate the open reusable data from the sensitive data. Users do not have access to sensitive data. It’s pretty well separated.
Q3 – James Reid) Have you actually road tested this with epidemiological data, at the coal face so to speak.
A3) Only tested within the university so far. Not publicly tested yet. But we are planning to make it as a normal Drupal module reusable in any Drupal site.
“Primary care perspectives on open source GIS in the UK”- Edgar Samarasundera et al, Imperial College, UK
I’ll be talking about how Open Source could be creating new tool sets that could be important to primary care. Then I’ll go on to look at some of the current GIS tools that are there for Primary Care. Then I will look at some of the gaps and how open source could fill in some of those gaps.
As many of you will know Primary Care has becoming increasingly local. Primary Care used to be delivered as part of large health authorities. They are more or less the size of a local authority now, more local in nature. And the current administration is looking at creating even more local authorities.
Correction from the audience: this is England and Wales not neccassarily Scotland where it is set up differently
There has been an increasing drive towards more localised geographic information. Over the last ten years or so we are aware of the growth of the internet and the growth of internet-based GIS tools. All of this feeds into wider digital healh information strategy for transparent information and analytic methodologies. And that ties in neatly to the open source agenda, moving away from proprietary tools etc. And there are other data sets – like location of fast food vendors say – that could usefully have applications in understanding public health.
So we have some examples here – a map visualisation around smoking cessation services and smoker density though this has been created in a proprietary ARC GIS tool. And this has helped the local authority in Nottingham find
And another example has looked at need for Primary Health Care vs. location of PHC providers – looking for gaps where there is need but low provision
Most recent initiative is the NHS Atlas of variation in healthcare – there is something similar in the US called the Dartmouth Variation in Healthcare virus. This produces static maps at sub-regional and regional level but looking at hospital admissions, not really PHC data.
A firm called Dotted Eyes has a health portal at PCT or ward level. People can pull off reports and maps for their areas. These are quantative but routine data sets.
There are the public health observatory interactive atlases. These let you look at a name, produce graphical information. It’s more interactive data and visualisations. But again routine data.
Something richer but purely PHC aimed are iQ HealthMaps which is a commercial tool (from Experian) aimed at PHC used at strategic level, PCTs, newly emerging GP-led consortia, individual practices, There is a seamless inerface with NHS N3 network – this goes down to the individual level which is why it is targeted at PHC. This data can be fed in Google Maps and also in Mosaic (also Experian owned) but using concepts and data that open source tools could be using. We are talking about qualitative data level. So areas of large amounts of fast food outlets, or higher level of smokers in the socio economic group etc. It’s qualitative and it’s different. Of course a good question here is whether this leads GPs thiniking pr the other way around.
Another tool Dr Foster Population Care Manager – this projects future healthcare needs, future morbidity, long term high end data.
But from all of this there are real gaps. There aren’t any analytics around
Spatial analytic outputs for surveillance and planning. Collaborative opportunity for academia. Open Source offers a cost –neutral option standardised to allow reuse of data, use of OpenLayers with multiple data sets etc.
A theoretical open source GIS tool
– information-driven tools – starting with PHC information rather than design tool
– Need interactive graphs, maps, data, visualisatiojns etc.
For example a map of unrecorded stroke prevalence – we’ve created epidemiological desease rates for the country – a model of predicted prevalence. And we’ve compared the outcomes from PH from the Quality Output Framework (QOF).
From a technical point of you we want R and OpenGeoDa links with QGIS – see slide for more.
So if we look at these patterns it can be hard to see what is a real cluster and what isn’t so we really want to run proper statistical cluster tests – so you can see real clusters in East London, including Tower Hamlets which is well known as a difficult area for PH to manage.
Here is another example where we use a larger kernal (5km) to compare the robustness of the clusters.
Another example here looking at Primary Care Quality Factors and Stroke/Stroke management with hospital admissions for Stroke as an outcome metric. And we can take this further with more complex data. We can overlay the data from the primary care factors, hospital admissions and the ratio between the two models. So here we have an admission with population factors as a predictor to hospital admissions – in Tyne and Wear the population factors is a good predictor, in Tower Hamlets by contrast the primary care data is a good predictor.
This could be done by PHC providers, local authorities etc. and using open source
Putting together a bid for the NHS with various web service funcationality, We are putting together a bid between Imperial, Notthingham and Heron (a pharmaceutical consultant) – being Open Source does not preclude that involvement of a private to the NHS. And we hope to work with primary health care providers, health observatories etc.
Q) I’ve been involved in a project called Smart Cities – looking at teenage pregnancies for instance – using Mosaic. Have you thought about those public authority local data sets that may be available?
A1) Not looked at that yet. But we have started to think about that, like teenage obesity etc. Hopefully some of those data sources and projects will come up in the discussion later
Q2 ) How have you found access to decent data, particularly NHS data?
A2) From our perspective Imperical College has a huge medical department so we have a huge dump of medical data of all sorts, as long as it’s not confidential we have it. I am a geographer but I work
Micheal Solyack sits on various NHS committees and contacts. From the outside you really need to collaborate with those that do have access already.
We are talking about analytic outputs, not data so open sourcing that secondary derived data should be fine. The original data would be problematic though.
Q3) the iQ Health maps is a tool that they sell. Who else buys this data?
A3) Practices I believe. Even before this latest move to consortia. iQ Health maps do market themselves and they have been around giving presentations to find customers. Tehre is a market for these tools but why not create open source tools that will be cost neutral for the NHS
Q4) I am quite interested in the possibilities for misinterpretation of the data. Different techniques can cause different interpretations of the data
A4) We would have an open description for the methods of interpretation, document those, have them as associated information to go with these maps. And something I’ve done as a pilot study at UCL we’ve worked on map data – it’s quite varied interpretations and misinterpretations, particularly for those less experienced with mapped data – particularly if no supplementary data sets, how the data has been produced. Many of the atlases lack that supplementary information so we clearly need to create supplementary notes etc.
Q5) How do you interact with these maps?
A5) With the public health observatory maps you can click on an area and view tabulated results and maps
Q5) We have some public authority data for Scotland with slideers for multiple types of data. You seem to be doing this just for two factors. Looking at our data you can see clear correlations between health and poverty in Scotland for instance – tools that can let you play around with the data in lots of different ways
A5) Yes, it would be advantageous to include that other type of data albeit with many sorts of supplementary explanatory data to make it clear how it works but it’s a good idea.
Q5) Multi touch is ideal for this
Comment) Do the NHS use multi-touch screens? You have to have tools that suit your audience. There is a whole body of literature about the geospatial literacy of those who are not experts in the field. Less is more and you should create tools that are easy to use
A) This is perhaps a matter of creating decision making tools that have more functionality as a more detailed level behind more public levels – tiered tools and points of access to this data.
Q6) How much is this a visualisation tool and how much is it a decision support tool? The nieve user can make conclusions the evidance doesn’t support.
A7) It’s a genuine issue with all these kinds of toolkits. We have statisticians and decision makers at one end but it would be good to have some information publicly available but it’s a really tricky area.
Comment – JR) There is a Open Area Classification which is for demographic data. It’s based at the University of Sheffield – there are some open health data out there. We also have a repository for geo data, not specifically for health data, called ShareGeo – we’ve just added data on locations of prescribing and non prescribing health sites. So there are tools for sharing data and information here in open source ways.
A) There are some distinctions between open source and open data but there is a strong relationship there. There’s a whole research project to be done there about what data you can and should release to the public and what the issues around spatial literacy here.
Speaker 3: “Green Space and Mental Well-being: Does Green Space Make a Difference?’ by Catharine Ward Thompson, OPENspace Research Centre, Edinburgh and Heriot-Watt Universities.
You can see I’m a landscape architect at the OPENspace Research Centre. I work with others who are not landscape architects. I’ll be trying to give you some information on some of the research projects we’ve been working on.
The simple answer to my title here is “yes” but we’ll go into more detail.
Research Context: green space in England – this research for CABE was building upon the findings from the Urban green nation: building the information base (HWU). This showed that the quantity of public green space in urban deprived areas is generally worse than affluent areas. And that Black and Minority Ethnicities tend to have less access to good quality green spaces.
We were asked to look at the quality of urban green space in six deprived areas in England, and to look at specific ethnic communities’ access to urban green spaces. We went to areas with high concentrations of black and minority ethnic populations and we looked for pair areas where similar levels of green space were availoble but where one area was high quality, one of lower quality. We did find that quality makes a difference. We used a conjoint analysis questionnaire to look at what environmental factors, including green spaces, make a good place to live. Safety and security were most important. House/flat suitability was 1.5 times as mportant as acess to green space. Availability of public transport is also important. Green space contributes about 10% to making the area a good place. Ethnicity was related to having more concern with safety. And these results can, interestingly, be mapped against broader results. 10% importance seems to be quite a common factor.
We found difference between ethnic cgroups on some key areas. Satisfaction with local neighbourhoose, satisfaction with green space, safety of green space. We could map those results spatially. And we were able to start to establish some national geospatial data that quality of green space matters, and that it differs. And that ethnicity and cultural expectations was a better prediction than income.
Another project we are involved with in Scotland at the moment, with the James Hutton Institute in Glasgow, is Evidence from GreenHealth. You may be aware of work by Mitchell and Popham (208) showing that if you have more urban green space near where you live then you will have better health for those living in povery/deprivation. de Vries et al 2003 found that the impact of this is greater for those spending more time at home. Often this research is to do with self-reported measures of health. And we are still not sure what the impact of green space is on health, we want to explore this more.
We wanted to look at cortisol (via saliva testing) as a biomarker for stress to see if that is associated with the amount of green space near the home. We did a small pilot study to see what the issues might be with recruitment, with the use of this marker, etc. We picked several areas but we took out high rise areas as that has separate issues and access to green space changes significantly with high rises.
Our sample was poor, mainly out of work, and mainly finding it hard to cope with their current income.
% of green space in each participants residential environment based on data from Centre for Research on Environment Society and Health (CRESH) (Richardson & Mitchell 2010, Mitchell et al, 2011) and from Ordnance Survey MasterMap. Our initial sample had differing green space levels.
We did a number of analysis. Cortisol patterns tend to start high and drop sharply down during the day – that’s a healthy pattern. We can see unhealthy stress levels where cortisol patterns start lower and do not drop off as sharply during the day. We looked at cortisol mean, cortisol slope, wellbeing, visual access to green space etc.
When we look at predictors to cortisol slopes we found that high green space people have a higher cortisol slope than those without. So the answer here from this small pilot study is that the difference between high and low green space does seem to effect cortisol levels. We are now conducting a wider study, looking at sub-group variations, looking at seasonality and recruiting door to door (the pilot was through the job centre).
There is a green space and gender interaction when we look at % green space and mean cortisol levels. The higher green space and lower green space levels on mean cortisol level varies for men and for women so we need to look at what’s going on there.
Looking at graphing of cortisol slopes high green space leads to a higher cortisol slope for men, for women the slope is higher but the levels of cortisols are generally higher for high green space vs, the cortisol levels for men. So there is a positive contribution from high green space
We think we have an objective measure of the levels of green space in he residential environment and we have found that levels of this type of green space can dignificantly predict self-reported stress and cortisol – a biomarker for stress – in urban deprived area. There is a gender effect here as well.
Q1 – Catherine) Is there a seasonal effect?
A1) Our pilot was in January, our follow up was in May. Our data was sufficiently statistically similar that they could be combined so although it seems intuitively like there should be a difference statistically it looks like there is not.
Q2 – Anand) C
A2) Our colleagues at James Hutton are looking at a lot more spatial data including ownership of spaces. We do have a questionnaire about perceived quality of environment and some work about quality that our colleagues will be looking at. It’s an interesting issue and we will see if we can look at that aspect of the data. Who owns the land may not matter but conceptually
Xomment) the Mitchell Poppert study mentioned earlier was a very general measure of green space – not on ownership or access. Purely about how green the sapce is. But that’s where we should go next.
A2) Even with crude objective measures there is some sort of relationship and that’s important here
Q3) What was response rate? And how do you measure cortisol?
A3) That was about 40 – 50% participation. For health work that’s low but for environment that’s a high result rate. Cortisol is measured with a cotton swap, put it in a bag and keep it in a fridge. We asked participants to tell us when they wake up and we sent a text message to remind them at that time. We took samples over 2 days. Those samples are relatively robust even if not refridgerated and we picked up those
Q4) You said you were recruiting volunteers. There is an app on current states of happiness that UCl are developing. It would be relatively trivial to gather wellbeing and location via mobile devices
A4) We haven’t done anything on this yet but the technology changes so quicvkly here. I am working on another project which started in 2007 using accelorometers to look at the mobility of older people. That was high end tech then and GPS was too tricky and bulky. Now you can easily have all that stuff in your pocket. That makes a huge challenge for researchers – to work out how to gather and use all this wealth of data that you can now collect?
Q5) How much need is there for an open green space/open quality of green space data set? MasterMap is restricted of course but there are open tools. Are there tools to further that open green agenda?
A5) Yes, there are so many ways to define green space, and different ways to measure access etc. The CRESH data on green space is available – that version of measuring green space is available even though based on data that was not. We want different types and kinds of green space data available, and commentary on this as well.
Comment) There is a project from Scottish Government which is looking at creating coordinated data set for green space – we are looking at how to manage and sustain that. However there is an issue with making it “freely available” where our attempts to make it available to citizens through a web service has been refused by our colleagues as we’d be expected to pay commercial rates for using their data. Once again we founder on the rocks of our relationship with our mapping agency.
Comment – James) Just to say that for academics for non commercial purposes you do have access to that sort of priviledged data
Q6) Did you distinguish between those with and without gardens?
A6) There was a question on our survey about that but it didn’t seem to make a significant difference though I think there may be a relationship between men, wellbeing, and access to gardens – I conceptualise that as men in sheds – but broadly across the population it wasn’t significant.
Q7) You talked about apps etc. What percentage of the population actually have smart phones?
A7) Our sample were aged 35-55 and they did all have mobile phones but not smartphones. But it is changing. For 65 years plus work that sort of technology wouldn’t be appropriate but may be in a short while.
Why are GIS/spatial tools under utilised in the NHS? led by Kate Jones, JISC G3 Project (@spatialK8 #jiscg3)
As James said I am part of the geospatial stream that JISC have recently funded.
My PhD looked at modelling social marketing and preventative health . We’ve also recently been looking at usability and learnability of GIS solutions. It’s very important to create useful and usable applications – they must be easy to learn and user friendly.
I did some work with Experian and was able to use Primary Care type data and was able to link that to a national geo-demographic application. We were able to model where likely admissions would come from, looking at where disease (e.g. diabetes) was not being managed as well – very important for practice.
Because these are PostCode profiles you can look a risk by neighbourhood profiles. There’s a lot of deprivation in Camden for instance – you can just think that only deprived communities are at risk but we geocoded the Health Survey for England looking at health behaviours (obesity, smoking, alchohol) and actually these map to young transient communities. This is a real challenge for social marketing as the community is continuously changing. When I was at Camden I was told that it included the largest social divide in the country. We were able to use OpenStreetMap data in this urban environment and look at the Index of Multiple Deprivation and to look at where the connections in the community are/how well integrated a community is to it’s local communities. So I’m thinking about using
Manifold GIS ($250) much cheaper than Map Info (most common NHS mapping tool). But you could repeat the same thing witH Quantum (an open tool)
Impetus fro this workshop – will the NHS lose it’s way on open data? And will it lose it’s way with geo? When I was in the NHS the mapping work was all based in the public health department – these are moving into the local councils so there will be a loss of expertise from the NHS. There will be a lack of GUS across NHS. Much discussions about OpenStreetMap – usable in some places but arguably not perfect for all locations. The GP practices hold a wealth of information that the PHT couldn’t access so with a focus on localisation this data could become even more inaccessible.
Observations. There are concerns about costs and budgets; data availability; skills particular technical literacy – Experian and Doctor Foster are probably tapping into that niche that needs these tools but lack any GIS skills; lack of GIS strategy – it’s not business critical in the NHS so how do you encourage the NHS to embrace the technology that is a support function only – there is a need for GIS evangelists.
Why I am interested in geo and health and open solutions? Well in the day to day activities of the NHS it is hugely underused. So first question – where is GIS in the technology adoption curve? (Higgs and Goulson’s work over ten years ago showed how GIS could be useful for NHS) And what are the barriers here? (what are the quick wins and the long term wins).
Comment – Annand) Barriers – I would put cultural high on that. When I speak to doctors on GIS or technology in general I use the term “learned helplessness” – it is the phenomenon of a professional group that has come to the belief that an area of expertise is someone else’s domain. My perception is that we are near the left of that curve. My feeling is that commercial organisations have been a lot smarter than the public sector. A side effect is that there is a generation of professionals who believe that technological solutions are brought from companies that provide solutions. That approach would not be acceptable to anything considered a core to the NHS but somehow information technology is seen that way
Kate) Before working in GIS I worked in the city and technology was seen as something to outsource in the same way. When I did work within the NHS I wanted to use something that was not a Microsoft
I even used Overhead slides to simulate map layers
Anand) You find the situation in Primary and Secondary care where the choice of tools sits with the IT department and they decide the type and choice of tools to be supplied. I’ve had very little success persuading health professionals that that is an odd way to proceed
Comment – James) Some of those issues with health and NHS are more systemic than that. We have representatives here from Scotland and the Geographic Information Sector in Scotland. This is an issue that is not specific just to health, it’s how you make geospatial information relavant to people who do not see this as a core part of what they do. The likes of Tesco know how this can help their business – why do so many other areas not see geospatial information as useful?
Comment – Edgar) In many other countries Primary Health Geo modules are commonplace but not here.
Comment – James) What is the modern John Snow here?
Kate) The crisis mapping
Comment – Cameron Easton, Scottish Government) I’m here to listen more than talk as this is very much a live topic for us. We do have a GI strategy, a spatial data infrastructure and we are prioritising geo spatial information. With one exception our health boards and special health boards are signed up to our Ordnance Survey service – at the moment it doesn’t cost anything but when that agreement finishs in 2013 we need to be in a position where rather than be in the chasm we are at least in the that peak in the middle of the adoption curve – so that geo is seen as a core part of business in those health areas. How do we as a public sector community promote and encourage and stimulate the movement to the right of that adoption image. Another bit of the bad news is that NHS Scotland is no different to other parts of the public sector – there is no one entity but a whole raft of different bits that could use GIS for different things – disaster planning, facilities management for instance. Facilities management has been rolled out in Scotland – that’s one little bit of use of GIS though a multimillion NHS project. Getting my head around all parts of the NHS that might need to work with GIS is difficult enough.
Comment – Derek Hoy) I’ve been working with an NHS funded project called ALISS. We have been looking at healthcare in the NHS but there is a whole world of healthcare outside of the NHS – long term health issues outside the NHS – we are finding the NHS encouraging individuals to take more responsibility for their own care. We are looking at how people map out the health related resoureces they use and find useful and putting them on the virtual map if you like, things that are local to them. But it’s different to putting a pin on the map – it’s about access to resources. when you get into virtual resources the map has less meaning. Some are in-person services in a fixed point, some are services that come to the home, some are virtual resources.
Comment – Bob Kerr, OpenStreetMap) Some of the resources that can appear on the map – we are working on OpenGL to let you place services on a map but you can visualise the physical parts of your database on a map and move between that and your database.
Comment – James Reid) How do we move beyond the map? How do we move from theoretical ideas to a place where you can reach the policy makers, the funders, the decision makers?
Comment – Bob Kerr, OpenStreetMap) In Craigmiller, a deprived part of Edinburgh, we made a general map of the area. What was interesting was that those in this area didn’t neccassarily know where resources were in the community. There are various community projects so we started mapping these out.
Comment – Annand) Bob mapped these and handed out A0 prints in doctors surgeries etc. People were really interested – they weren’t used to seeing maps of their areas and they were discovering things that were local but they hadn’t known about before. The presence of a big quality colourful paper map was really useful. Also things like walking paper
Comment) Some of the quick wins – if you want to release a new dataset you have to go through an ethics committee that meets once a month. A wait of perhaps 3 months just to get a proposal seen. Organisationally you have IT to deal with, also senior advisors and consultants (Cap Gemini etc) – they want the best in general but can take time. So streamlining that process would be hugely helpful and having an agreed process to anonomise data in real time – SkyStore should already be able to do this – and release it. The big barriers is approval not skills. Having data available is crucial.
Comment – Anand) I’m not sure SkyStore can do that. And I’m not convinced we have a standardised way to share health data.
Comment – James) Cameron mentioned the spatial data infrastructure and that fits with the European INSPIRE directive. But there is a real disjoin between the top and bottom
Kate) I was really pleased when the Home Office released crime data at street level. You can’t perhaps release data at that level for health but you can share statistical data in some detail.
Comment) If you can find data on a single person you have failed to create a good database – solve that and opening data becomes easier
Comment – James) I’m aware
Generation Y seems to have a different attitude to the sharing of personal data – you see demographic and health data in Facebook that traditionally we’d see as so off the wall as to be unusable. There is a commercial imperative to mine this
A university in the US have done some work in Twitter
Blue Denghi epidemiology studies in Brazil
David Banyon, Prof of HCI here) Most of the discussion here around the data not the interaction. Make it sexy, make it fun and you can really make a difference to interpretation of that data. Come to room C78 in the lunchbreak where we have a multitouch enabled room with GIS data.
Kate) Gapminder uses Google software lets healthcare providers look at key statistics. The academic behind that, Hans Roslin, helps make this stuff really sexy and appealing and visual.
Comment – James) When I was a student we talked about “how to lie with maps” and perhaps we need to change “maps” to “GIS”. We need to be clear about our objective in the things we build. Presumably the underlying objective is improving the underlying fairness fo health
Comment) Actually if people within the NHS could see what you are doing they might have different perspectives to us. Something missing from the list is dissemination – who are the stakeholders, how are you disseminating this? So important to get NHS and healthcare providers into an event like this is so important.
Comment – James) Health geographies is one of the most notoriously dynamic and it’s so diffuse and multifaceted that it can be hard to engage those health providers. There is a legacy there and it’s difficult to change directions in short order. It’s very hard to get NHS people along to these sorts of events…
Comment – Bob Kerr, OpenStreetMap) I gave a talk to doctors recently and they loved that. I showed them the map of Haiti and they needed some encouragement to participate
Comment – James) Maybe it really is about promotion
Comment – Bob) Just having these types of conversations is so important to start these things off
Comment – Bill Buchanon) To me that is so important. There are some stakeholders here. We have to engage them. Two statements just then are wrong: it’s not difficult to get healthcare providers interested but you have to find the agenda – assisted living is a current concern for instance. The other was that GIS doesn’t save lives. That’s not true. We did a trial at Imperial looking at the flow of patients moving around using GPS and we found that adding one person to the radiology department you could improve the throughput – you can make business cases here.
Comment – James) We do often bundle geospatial data and GIS here and that can cause problems. Locational information can be lifesaving. When you brand it into the GIS world it’s harder to sell. Movng to location information and real time use can be really helpful here for showing the relevance of geospatial information. A Queens Award winner recently put GPS in an insulin injector – seems no brainer but needed innovation
Comment – Serge) Lets assume that health data is out there and on the web and can be used. The Home Office mapped crime levels in different areas. Anyone can open the site and check the property and the surrounding areas. After half a year an observation was that people stopped reporting crime as they didn’t want their homes on the map. If we did this with health imagine what these unintended consequences could be.
Comment – David Hoy) BCS has a health Scotland group and Angus and I are on the committee. There is an event in a few weeks – there are presentation and workshop opportunities and the group there includes a number of NHS people.
Kate – When I thought about this there were 3 or more projects in this geospatial stream that could help the NHS. G3 is helping look at how we can teach non specialists about geospatial – see also the ELO GEO and GEMMA projects.
Body storming the subject with GPS by Chris Speed, Edinburgh College of Art – Bring an iPhone if you can!
My background is as a digital artist and my interest is in how code changes the way we behave. It’s tangental but it is related to health – changing behaviour through connection and obsession with these devices.
We actually worked on a project called Walking Through Time that got some 9000 downloads in the 4 weeks of the Festival last year. It combines historic data and maps – there’s a blue dot – and that’s you. If you see that dot on a cattle market and you look up and you are in Grassmarket then you are disrupted and learn something! So, I’m here to disrupt the day!
GPS pops up in domestic areas – Satnav have changed how we see streets – most is offline but increasingly they are live through simcards/mobile phones. They tend to do one thing and one thing well – route from A to B. Hopefully in the right way. Android and iPhone apps do the same thing but you have to root around to find other things to do with GPS.
The output of GPS devices can be exported – you can use or write apps to see your trails. A lot of GPS now will take you down to 7 metres or 5 metres… that’s enough to put you on a road or a pavement. And you can draw with your trails. Jeremy Woods has done this. And Daniel Belasko Rogers has been using GPS for years so he’s created his personal map of Berlin. OpenStreetMap has long been doing this – here we can see video of a mapping party – but you don’t see these live maps so often.
Christian Knowles work at Biomapping.net used skin sensors on walkers to biomap stress as an indication of traffic. Of course there are problems – if your girlfriend rang and dumped you you’d get an unrepresentative peak of stress in a random point on a map.
Soft GIS. You get quite precious about your device, it’s very personal and trusted and private. Mappiness is that app James mentioned which pings you once or twice a day to check your happiness with a 1 minute questionnaire, location, and who you were hanging out with. It’s a PhD student project and there are big caveats with that sort of data.
We wanted to do something disruptive and that’s where Comob comes in. You can create a group of people, moving through the street in patterns thinking about different ways to move through the city. So we are going to form pairs and become one node in a mob. We’ll go see this mob moving around and quickly you’ll find a particular location – you find that we don’t share conceptual models of, e.g. pollution. We found that poeple mapping a concept in real space and time and that triggers interesting conversations. Not much to do with mapping but mapping in situ is quite a powerful way to start a conversation.
We have a projection piece of software that lets us review movements – replay and talk through experiences. And that’s great to do but we’ve had a wee problem with our server so we won’t be able to do that today. So I’m going to ask you to go find me some health! Greenspace maybe? Something subjective! We will all have different ideas.
So… welcome to the app! We have a map view, a settings button, and you have a button to zoom to group. You can also use a pinching action to move in and out to zoom. The nice glowing ball – that’s you! When we move outside and away from each other we’ll make a shape – don’t worry if those dots look weird and messy now. You should have a Mark button – press it now! It should go red. That helps you communicate that you are happy with your status. That’s the hard stuff…
To the settings screen. We should decide on our mobs. lets do 2 mobs – mob 1 and mob 2. Then pick a username for your couple! (and switch on “show names”). And that’s that…
Finally put on Conv. (Convex) Hull. This will draw lines around the blobs on the map.
Cue much faffing with app set up! Now, go forth and form crazy shapes outside. When you are in position stay still and press Mark and wait until all in the group have marked that shape.
Final Workshop: Mini presentations from Bill Buchanan, Edinburgh Napier University on infection tracking and and future ideas, and from Seraphim Alvanides, Northumbria University on his research into the relationship between built, natural and social environments, activity and health. This will be followed by discussions of ideas and possible project building.
We seem to have lost 1/2 the group on a mapping exercise. Chris relays some of the issues about using data collection with the cloud. Examples of the iranian rising and the issues that may come out of recent riots in london. Will blackberry pass over the details of the ring leaders? Time will tell.
Martin Graham, Napier University
I’m a researcher in data visualisation and I have been working on some data from the Scottish Government on areas of multiple deprivation. This is an exploratory interactive sort of demo. This is the council wards in Scotland, ordered according to barious bits of information. What this is is a series of linked visualisations that shows how the data interact. On the left we have an outline of Scotland. Every council log shows as a zig zag line. We have a slider to show attributes and you can zoom in to find these data on the map outline. You can see the correlations on the map but of course you can’t see causation here, just that there is a relationship. You can try and filter some of these to see high income and low health or vice versa. And you find little wards on the map. We’ve used this on various data sets already. Colin Colm developed the GIS for us.
Q1) Is it available for us to use?
A1) Probably not, this is just a demo we have on the website. The tool takes CSV files.
Q2) It would be good to have a cartogram…
A2) We did have a version where it automatically zoomed in to areas of interest
Q3) Have you approached or been approached by local authorities? Given that we’re into the next period of census data and planning is foremore in most councils mind
A3) My boss is looking at possible use with census data. This data set was given to us by the Scottish Government. We’ve tried it out on various data sets to date. The speed on the interaction is constrained by the number of records – probably up to 80k records with about 20 attributes.
Public Health and the Environment – Dr Serephin Alvanides
This will be a short presentation. Two case studies from the North-East of England. The first is on obesogenic environments. We came together as a team to look at what makes an obesogenic environment, how we measured these and some data issues. So, BMI is expanding, we are getting bigger and the UK seems to have an increasingly percentage of obese people. There are all sorts of factors here – cultural, genetic, behavioural but we are interesting in environmental factors. Something to do with availability of food and activity levels – and it does link to green space. So in terms of defining factors the Foresight report of 2007 came up with a plethora of factors. I am particularly interested in food availability. There is a lot of research, in Scotland in particular, on food and fast food availability.
In terms of measuring the obesogenic environment we have factors that are food related; those that are activity related – condition of environment, transport availability; geographies – loved the example of forming our group geographies earlier, these aren’t always official geographies; methods – various available here.
Access and availabilities of facilities (e.g. food) you might look at the buffer around the home, the local area. It depends on what you would want to look at. You can look at the area level and all the potential types and locations of food outlets. And then you can aggregate individual factors to find the wider obesogenic environment. And you can have have pseudo-individual level – this comes with an awful lot of assumptions.
We think of maps like this visualisation of MacDonalds restaurants – but the clusters also relate to urban areas – it maps people as well as outlets. Then mobile markets in Greece or Japan change over time – how do you map that availability?
In terms of activity related environments you can use similar methods to map these.
We decided to explore the concept of walkability – residential density; connectivity; land use mix; deprivation – these are standard terms in the literature. See my recent Health & Place paper on this (Burgoine, Alvanides and Lake (2011). Assessing the obesogenic environment of North East England. In Health & Place, 17 (3). Pp. 738-747. DOI: 10.1016/j.healthplace.2011.01.011.).
). We came up with a walkability map of the North-East with dark areas for rural areas, yellow are most walkabilities. This tends to reflect the urban/rural divide.
The point is that if you try to do correlations between these areas you tend to see high correlations with those aspects. You can compare aerial shots of these areas to see how those trends look in terms of real use.
What do you do with this data? It’s hard to do this sort of work. We wanted to use in combination with the Health Survey of England (HSE) to validate the accuracy of the indices, examine the effect of environmental factors on walkability and in fact we are still analysing this data at the moment.
But there are other factors here – active travel includes cyclability as well as walkability – everyone is trying to encourage this. How do we raise Grea Britains 2% active transport to something nearer 10-20% of the Netherlands say.
Godwin Yeboah at Northumbria is looking at to what extent the built environment supports or constraints active transport-focusing on cycling in Newcastle-Gateshead using tracking, diaries, etc. We found a project called bikability.dk which is looking at similar issues in Denmark.
Q1 – Bill) What could make Edinburgh a more active city – there is potential
A1) Edinburgh has 6% cycling, it’s above the tipping point
A1 – Anand) The council’s move for 20 miles per hour speed limits will be a big help
A1 – Seraphim) Newcastle has also done this.
Comment – Addy) Those countries with high uptake are very flat…
Cue a big row about weather, topography etc. Mixed feelings in the room
Q2) Could you scale this up nationally
A2) Tom Burgoine, Newcastle University is looking to create national data and it may well be made available.
STEM – AngusMcGann
Spatio-Temporial Epidemiological Modeller. This is an open source tool. Various specific epidemiological models are already built into the tool and you can also include transport routes, migration patterns, interventions etc. It has been used in Mexico in the last flu outbreak and it has also been used in Israel and Palestine recently. For more information have a look at the GECO blog post my colleague recently posted.
Conclusions on the day and thoughts on future follow up
How can we better connect academic output, like the work we’ve seen today, to policy makers
Bill) If we can come out of this event with definite tangible things that would be super. A lot of funding looks for tangible knowledge exchange patterns. And there are particular interests in assisted living – systems around the individual and privacy issues around that as well. Would be great to have events with SMEs from Scotland and the North-East of England that would be great and involvement of stakeholders in these sorts of disussions seem important
James) Seraphim mentioned the Secure Data Service – this allows anonomised data to be used easily but it can be a really difficult area. Health data tends to be treated as extra sacrisanct. The more generic question is perhaps do we have the balance of access and confidentiality about right?
Anand) By default the concept of confidentiality of all patient data is that it should be private outside of the consultation. That is proper but that is specific to one particular context. It’s not neccassarily sensible to apply that confidentiality to all contexts. Actually we can think our assumptions about confidentiality in this society are seen as universal when they are not. It needs to be discussed bravely and that needs to happen before we start sharing data out. This has to happen in the community, not owned by the profession. The constraints of the profession are fine but the community has to have the deciding say in what is and is not confidential data.Part of the bravery is facing up to what is happening at the moment. Some people readily publish confidential information about themselves, friends, family on social networking channels. On the other hand we have people who are not comfortable with that but may have data about them shared by others – from the state and not of the state – and part of the bravery is admitting and examining where we are in terms of confidentiality.
Bill) We’ve tried to create a new platform that allows people to share data about themselves – too often the healthcare provider thinks that – totally private; shared with healthcare; shared for research purposes – we applied to the ESRC to create cloud environments inside hospitals and could filter out anonimised research data beyond the hospital. We had great support at Westminster, great support from the BMJ but the Daily Mail got hold of it and kicked up a storm. So it hasn’t happened yet. If we could create an open platform for sharing this sort of data for academic research that would be a great start. We need to educate people better about the cloud here
James) JISC has very much been moving from default closed to default open positions on data/tools funded. Research councils also have been doing these. That can shift when you look at audiences beyond academia. Nigel Shadbolt and others want to push that to
Edinburgh University is building a private cloud. There is a JISC cloud being developed. Some of these issues can’t be fixed by academia – we need wider awareness.
Bill) It needs a complete rethink!
Serge) The word security is overused, it’s a general excuse to not do things. We have to stop accepting this word. There are levels of security that can be established, we need to know why these things are not safe.
Ella) What Bill has highlighted is a problem that is public engagement – that’s one of the biggest issues for academics but of course it’s a huge issue for the health service. They have been trying to transmit messages about public health for decades.
James) How do we do that concretely?
Anand) I really loved Kate Jones’ idea of visualisations for social marketing – that seems like a useful route.
Bob) Patients Like Me is a site for patients with degenerative diseases. The patients volunteer to keep diaries of their health and what they take. It’s a crowd sourcing model. That’s an interesting model for collecting and sharing data about health.
Bill) Assisted Living is the right arena for that sort of debate and model.
James) Is there a sense that this is a Scottish thing? or a UK thing?
Bill) We spent a long time trying to engage with the NHS and government in Scotland and it never happened. In England we haven’t had that problem. The financial situation is not quite as bad in Scotland – perhaps England is more keen to find solutions right now. I think we need to get the NHS involved in this and get them to set the agenda.
James) Can we pressure that, can we say what the agenda is? We can set up a specific mailing list for the delegates for this event to stay in touch. We can do that as a consquence of this.
Caroline) We can also go to them – posters at NHS events for instance.
Ella) Could we write up this workshop as a letter for the BMJ. We spoke to a group of medical librarians recently as part of a project – they said that you should go into medical areas, medical events etc. They have one journal that they all use, lets use that target.
David) One of the problems with the NHS in Scotland is that it’s a bit anarchic. As soon as something is seen as being towards information and IT is not to have national projects. Speaking to people at the centre will get you no-where. They give money to initiatives at the health boards. In terms of epidemiological and GIS projects it would be more fruitful to approach the health boards. There are also some national agencies looking at health information for the public. You have to choose your target and focus on that and do some intelligence on where to approach.
Comment) Life expectectancy of the Health Protection Agency is at weeks not month but there is an event coming up soon that we could be at.
Bob) OpenStreetMap is having a conference at the end of this month – State of the Map Scotland – all are welcome. It’s a two day event and all are welcome. It’s in Glasgow.
Chris) We’ve found that we’ve got data sets from various places. Playful things open people up and help people get ideas. Especially stuff on their phones that let them embody and embrace this stuff. There may be some small projects that help nudge this along a bit. So a GP app that helped GPs get on board with GIS.
James) It’s ironic that none of the funded projects directly connect to health but we’ll take that message back.
Finally I would like to thank all of you for coming along today. To Bill for organising the venue and catering and to my colleagues Addy and Nicola.
We would love to have your feedback on the day (if you followed the liveblog only please add a comment to reflect this) – we have survey for your feedback and to join the email discussion list for GECO (we will email attendees about this too).