Integrating GIS and spatial statistical tools for the spatial analysis of health-related data.

Spatial Statistical Analysis (SSA) and Geographical Information Systems
(GIS) are instrumental in many areas of geographical study. However, their use
tends to be separate one from another. This has prevented their potential in many
application areas from being realised. This research is an attempt to bring the two
technologies together for a specific application area - health research. There are
two research objectives. The first and main objective is to construct a software
package - SAGE - by integrating necessary SSA techniques with ARC/INFO, a
GIS, to enable the user to undertake a coherent study of area-based health-related
data. The second objective is to evaluate and demonstrate SAGE through a case
study.
A range of SSA techniques was identified to be useful for addressing
typical health questions. A three-tier client-server model was suggested and
argued to be the most appropriate for integration as it takes advantages of both the
loose-coupling and close-coupling approaches. Under this model, a SSA
component forms the client, while ARCH/INFO functions as the server. They are
linked through the middle tier - the linking agent. The development of SAGE
provided experiences useful for developing a generic SSA module in the future
for any GIS that confonns to a set of well-defined standard application interfaces.
An empirical study of colorectal cancer (CRC) incidence for the city of
Sheffield using SAGE is presented. It shows the usefulness of the SAGE
regionalisation tool in constructing an appropriate regional framework for
subsequent data analyses and of both exploratory and confirmatory spatial data
analysis in exploring the characteristics of CRC incidence. Some weaknesses of
SAGE are identified, while remedies for them are suggested. Future work is
recommended.
The SAGE User Guide, related publications and the SAGE source and
executable code as well as the data used in the case study are enclosed for
reference.