Volunteered Geographic Information

Yesterday I attended an Infectious Disease Research Network (IDRN) course on “Use of mapping sofware and systems in health research“. The broad remit of this event was to draw attention to the use of mapping software, such as GIS, and geospatial technologies, such as GPS, in health related research. The event was chaired by Professor Graham Moon, who seemed interested in the subject matter despite the fact that his work is not strictly in infectious diseases, rather health-related behaviours.

The way in which i think about health geographies is very much related to the individual, and the condition of a person being healthy or not, the likelihood of certain groups having a certain level of morbidity etc. Thus the first few presentations were actually quite strange, coming from an infectious disease angle, the focus is not the person but the disease and how it spreads. Therefore several talks predominantly featured the preponderance not of people, but of frogs, larvae or snails. The upshot of such analysis is that mapping is really quite important, these creatures can only exist in certain conditions, temperatures, soil types, moisure levels etc. and so spatial analysis helps directly map their habitats and delineate the sites of highest risk.

This type of analysis coupled directly to the extensive Malaria Atlas Project (MAP) based in Oxford which trawls for contemporary research on malarial prevalence globally and uses the subsequent huge data source to inform stakeholders of where certain types of malaria are likely to exist. The MAP project uses an impressive set of workflows through PostgreSQL and Python to create nice maps (cairo, mapnik and reportlab) and web maps (Django (python), Java and google web toolkit). Their most interesting element is the bayesian modeling framework which takes malarial influences such as temperature, elevation and other environmental factors, creates a malarial range and is then able to say where you are likely to find certain malaria forms. All the work coming out of MAP is cross checked by a panel of experts.

In the first session there was also a talk from the guys at spatialepidemiology.net who clearly have an extensive knowledge-base set up. Most interestingly they are pushing the contribution of ‘citizen scientists’ through their Android-based software for collecting field data – called epicollect – however the extent to which they are actually engaging citizen- or ‘street-’ science is negligable. they seem to subscribe to the idea that collation of data from many sources counts, even though the ‘citizen scientists’ they envision are all fully trained epidemiological researchers, hardly ‘Joe Public’.

The other presentations came from the Health Protection Agency (HPA) who reopened the onrunning debate on patient identifiable data, dynamic and time conscious data and the rights and ownership of data ‘masehed-up’ in a google maps interface. Subsequently we learnt of the incredible work being done  by MapAction, who let us know that maps were also a form of aid like food, water and medicine.

In spite of this, the general sense of cartographic technique from the health mapping was weak excluding perhaps map action and the malaria maps. In general representation choices weren’t that intuitive, and there was an overreliance on points in webmapping. Further the almost ubiquitous use of binary red-green schemes is unintelligable for the colourblind. Generally cartographic design could have been improved for most projects. There were some interesting uses though, such as time-lapse tracks of women and children in Uganda who had been tracked using GPS.

Unfortunately (or perhaps fortunately) the open session provided some more interesting papers than the health-based session did. Particularly the fascinating work on the Legible London project, I was particularly taken with their treatment of neighbourhoods in London. The concurrent poster session provided some interesting comment, including excellent work on cycles and rhymes in the city from Urbantick Fabian Neuhaus, who also provides an extensive review of the talks on his blog.

Finally the keynote, Mikaela Keller involved in the Health Maps project, deserves a mention. The scope of the site, which is well worth visiting is incredible, the level of technical expertise that goes into the site in terms of natural language recognition and machine learning is enormous. Again though, the mapping seems to struggle with overpopulation of point sources, overwhelming the slippy map surface. Still, a valuable resource based upon collecting and displaying media reports of disease outbreaks from diverse news media and health agency websites – roughly 20,000 of them resulting in 200 new alerts per day.