300
1 The GeoMed Project
In 1998 the University of Michigan and soft
ware company BioMedware received a two-
year grant funded by the Small business Tech
nology Transfer Research (STTR) program of
the National Cancer Institute. The objectives
of this grant are
• to produce software for surveillance and
analysis of disease clusters, local spatial au
tocorrelation statistics, and related tech
niques (GeoMed), and
• to prepare and evaluate educational mod
ules for teaching spatial analytic theory and
methods applied to human health data, and
to use those modules in the classroom (in a
course we call “Spatial Epidemiology”).
We thus describe a work in progress: we have
offered the course once, and will offer it again
this winter term. In the meantime, we con
tinue to develop and (we hope) improve the
course modules and software. In the follow
ing we present the preliminary products of
our venture, and share with you the exciting
prospects for collaboration and distance edu
cation offered by our web site.
2 The Problems
What sort of problems do we treat with the
techniques and software of Project GeoMed?
One way to answer that question is to list
some of the projects our students undertook
in the last year:
• Kriging Illinois County Breast Cancer Mor
tality Rates
• Dengue fever virus antibody in Santa Clara,
Peru: Is there spatial autocorrelation?
• Spatial Clustering of Scleroderma in Three
Michigan Counties
• Spatial analysis of Anopheles [mosquito]
density in western Kenya
• Spatial analysis of Anopheles density in
western Kenya (dry season)
• Using Data from a Landsat TM Image in
Geospatial Analysis
• Exploratory Space-Time Analysis of Sub
stance Use Among Adolescent Students in
Puerto Rico
• Study of Perodontal Disease Symmetry
• Assessing Spatial Autocorrelation of Intrao
ral Loss of Periodontal Attachment
• Prevalence of antibody to an arenavirus
(probably Whitewater Arroyo virus) and
spatial and temporal patterns of infec
tion in populations of the white-throated
woodrat, Neotoma albigula, in southeastern
Colorado
• Characterizing the Transmission Dynamics
of Tuberculosis in Detroit, Michigan
Note some of the key words in the titles above:
• spatial autocorrelation,
• spatial clustering,
• exploratory analysis,
• space-time analysis,
• spatial and temporal patterns,
• kriging,
• transmission dynamics.
These words provide a fair coverage of the
materials we study. We emphasize techniques
which detect and characterize spatial autocor
relation, detect disease patterns, model dis
ease patterns, and aid in the determination
of the process behind particular disease pat
terns.
3 The Solutions
In outline form our web site solutions include
• Modules - self-contained units for instruc
tion or self-learning, including the following
components:
• Lecture material - the basics of the mate
rial discussed in class, in more or less de
tail. Some of the lecture material is anno
tated PowerPoint lectures; other lecture
materials were developed on the web; still
others are being converted to web mate
rials by scanning documents.
• Lab material - computer labs designed to
test the student on the theoretical mate
rial presented in lecture.
