April 03, 2008
Live Blogging from KSL GIS Symposium--Plenary Session #1 Charles H. King
Joseph Koonce, CWRU biology professor, introduced Charles H. King, MD, from the Center for Global Health and Diseases and Dept. of Epidemiology & Biostatistics, at Case Western Reserve University.
"Microsope to Macroscope - Using GIS to Understand Environmental Complexity in Dease Causation."
* Background on the philosophy of medical science
* Resistance to GIS
* Complexity and theory of environmental analysis
* GIS and the new practice of "eco-social epidemiology."
Since the 1700s the microscope has been the rationalist/positivist tool of choice; reductionistic; germ theory was a major breakthrough; now molecular medicine.
What is wrong with environmental studies in medicine: It's messy; It's too complicated; it's difficulty to isolate cause and effect; it tells us things we don't want to hear. Complexity reigns in our political world. Genetics, exposure and environment all relate to infection of disease.
Radomized control trials do not reflect the real world: "Why don't patients get getter on 'proven' regimens?'" There is "hidden stratification" in samples that will end up with unpredictable results.
Complex systems organize themselves into predictable but chaotic-appearing patterns.
What are our health research goals? Explanatory; Predictive (past performance does not necessarily predict future performance)
He discussed his own research about transmission of a parasite spread through water and snails in Africa. He described his use of GIS for analysis/data mining. Use of remote sensing and satellite imaging for mapping, creation of spatial data. Data must be confirmed on the ground with GPS data. Data is correlative, not causative.
Other dimensions such as poverty and socioeconomic factors play a role. We cannot ignore the context.
Why do we do this research? We want to be able to use all of these factors to analyze how they combine to foster disease.
Opportunities for GIS:
* New impetus for ecological research
* Comprehensive multi-scale picture of local/regional/global epidemiology
* Consideration of temporal changes
* Integration of molecular data with environmental.
Spatial data, concerns and limitations:
* massive amounts of data,
* but paucity of of accurate epidemiological data
* lack of data is readily masked in maps
* meaning of area boundaries is importans--show what you don't know.
* There are limites to the use of spatial auto correlation and interpolation
Q. How do you obtain data, and how much data is available?
A. "It's all over the place" Rainfall data for the city is discarded every day; some weather data is retained, but it may not be what is needed. It may be necessary to set up your own sensors, such as what Prof. King did in Africa.
Q. Comment: a focus on complex ecology is important. Multidisciplinary research (ecology + disease) is a problem for NSF and NIH--there is one joint panel to handle such applications.
Q. Can mapping move to prediction?
A. Some generalities can be made for some cases, but in most instances there is not enough data to make predictions (e.g. West Nile virus).
Posted by tdr at April 3, 2008 11:07 AM