Pediatric asthma epidemiology map
Posted 27 January 2011 - 05:09 AM
This map is part of my job as a regional epidemiology. The maps that I do not I need to be especially beautiful and attractive but above all must be well understood by doctors and health facilities. Despite the "normality" of the map shown here is hidden behind a long job of standardizing data, geocoding, statistical analysis, use of dispersion models.
I want to show how the pollution of two major roads (blue lines)have an effect on the distribution of the asthmatic disease . The purple patches are the Kernel Density of cases of disease that it can be seen along the lines of highways.
In the second image shows a model of critical traffic on local roads and the pink circle defines the standard ellipse that expresses both the degree and direction of dispersal.
The analysis was conducted on the territory of the Province of Verona, Italy.
I hope you enjoy my work which is an alternative way to use the GIS tools. Italy is not the best place to conduct such studies, due to a general lack of funds and investments in research, especially for a lack of trust in young students like me. Politicians are too busy trying to defend themselves by personal problems, as these days that see Berlusconi defend himself against the child prostitution. For this reason I hope that between you there is interest in my work and there is someone willing to offer some kind of opportunity.
Posted 27 January 2011 - 10:55 AM
I can understand it well enough, though I think that you are in need of a title and legend. Maps should always be stand alone. It is not just about making them pretty, it is about making them correctly.
My initial reaction is that these data are raw. Are you mapping the percentage of asthmatic cases versus population or just the number of asthmatic cases? Of course areas such as Peschiera, Verona, S. Martino, Soave, etc... are going to have higher number of cases b/c there is a larger population then say the Valpolicella region.
Hope that helps,
Posted 27 January 2011 - 11:38 AM
I'm also curious about the statistical analysis around the first maps display of disease distribution. The large purple area could correlate to large population core, more people equals more incidence of any particular aliment. Also what are the socio-economic factors? Residences right along major highways tend to be lower income. A lack of health care access may also be reflected in the distribution. What other possible factors are nearby? Is there a industrial zone along the other side of the highway? Showing where all other major pollution contributors are can help shed more light on the study of the highways as the main culprit.
When doing GIS analysis like this always remember that correlation does not equal causation. Because there is a geographic correlation between these two features (highway pollution and asthma distribution), does not mean there is a direct correlation. There may be many other factors that can be aded to the map or mapped separately to more fully explore the phenomena.
On the maps themselves, since you are mapping the same area with different analytical data, I would set a single base map design and only vary the statistical information's color and style. So keep the roads one color, keep the the projection and map size the same. Use the same back ground color or image. Label major locational features like the two main roads and the town center. Good luck!
Posted 27 January 2011 - 12:55 PM
Posted 27 January 2011 - 01:06 PM
With regard to the legend and the basic information you are absolutely right, I forgot to include them, but I was curious to expose my work and I forgot to include them.
The study was presented at a conference of planning last summer and found a lot of interest (http://www.slideshar...andro-seravalli (in italian)). In the study were taken into consideration:
- All children with at least one prescription medication for the treatment of asthma (cases)
- All children in care of the province divided by the municipality of residence (controls)
- Geocoding data collected from air monitoring stations in the province
- Data traffic on the main roads
- Details of the pollen season
During the project I realized that putting together such a great source of data to correlate it would be too complicated and would make the study vain.
I drew a map of the prevalence for common and I did a statistical analysis to find clusters but have not been identified.
Modeling that lets you see the dispersal of pollen is extremely complicated and can not alone be used, as well as PM10 and NOx. So I concentrated my efforts on the data traffic.
I run a buffer according to road traffic measured concentrations (sum of traffic from heavy vehicles and light) as seen from the map 2 and if it relates to the kernel density of the cases (map1) can be discerned report.
I'm about to graduate in Geographical Sciences and I have no training in statistical analysis so I had to make do as I could and could not do complicated calculations using only arcmap for all the work.
I think that as a first approach to the GIS tool is not bad, but surely the study is not completed and should be further developed with more appropriate analysis. I welcome suggestions.
Thanks to all
Posted 27 January 2011 - 04:30 PM
The study was presented at a conference of planning last summer and found a lot of interest (http://www.slideshar...andro-seravalli (in italian))
Interesting work, unfortunately I can't read Italian (as much as I wish I could) so I couldn't see about your methods. You might be interested in Waller and Gotway's work on spatial statistics, Applied Spatial Statistics for Public Health Data. They have a section about doing Kernel Denisty with cases and controls. My guess is the controls would be how you could answer some of the questions that the previous posters have brought up, particularly David's comment about the population density in the purple area (e.g. second order effects). Still, though, you got to start somewhere, and this is certainly a good start.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users