|
William Thilly, Ph.D.,
Professor of Biological Engineering and Director,
Center for Environmental Health Sciences, Massachusetts Institute of Technology
Summary: Data for cancer mortality has been assembled and organized
by birth cohort and age for the United States since 1900 at http://epidemiology.mit.edu.
These data suggest major increases in lifetime risk for cancers (in addition
to lung cancer and smoking) that began with the Industrial Revolution
in European-descended and with Emancipation for African-descended Americans.
A similar pattern obtains with the Meiji Restoration in Japan. (Ph.D.
thesis of Dr. Pablo Herrero-Jimenez)
In a parallel study the data for cities and towns since about 1958 have
been assembled, organized and analyzed with regard to testing the null
hypothesis that there are no significant differences in cancer mortality
rates among cities and towns for adult (ages 65-85) and pediatric cancer
mortality rates (ages 0-18) in the states of MA, PA, NY, CA, TX and FL.
The results are interesting and lead to questions about "cancer clusters"
and other forms of public health data used to ferret out environmental
risk factors. (Ph.D. thesis of Dr. Janice Vatland)
Biography: Bill Thilly earned his B.S. and Sc.D. degrees at MIT
in '67 and '71 joined its faculty in '72. He has been pursuing the mechanisms
of human germinal and somatic genetic changes and their roles in human
diseases ever since. His laboratory developed and patented human cell
point mutation assays, human cell mutation assays in cells expressing
human drug metabolizing enzymes, microcarriers for mass production of
viral vaccines and the technology, constant denaturing capillary electrophoresis
(CDCE) that permits identification and enumeration of rare point mutations
in DNA of human cells, tissues and pooled blood samples from large human
populations. While exploring the kinds and numbers of mutations in individuals
his group developed an interest in matching hypotheses about the roles
of genetic change and tissue cell kinetics with the observed lifetime
experience of populations. To this end a subgroup led by Dr. Pablo Herrero-Jimenez
created the first comprehensive data set for mortality in the United States
(1890-1997) making it available for researchers at http://epidemiology.mit.edu.
More recently his group, led by Dr. Janice A. Vatland, gathered, organized
and re-analyzed the cancer mortality data from ~ 1958 forward for individual
communities within the populous states MA, NY, PA, CA, TX and FL. From
these analyses come the questions his group has been asking about the
widely-held but untested hypothesis that there exist significant differences
in cancer mortality rates among the hundreds of communities that comprise
any of these states in the latter half of the 20th century
Bootstrap Sampling and Bootstrap Estimation
James Heltshe, Ph.D.
Professor of Statistics, University of Rhode Island.
Summary: I will explain the ideas behind computer intense methods
that are currently in vogue in applied statistics, particularly in the
environmental sciences. Computer intense methods are applicable in situations
where one does not want to make statistical assumptions about one's data.
These methods are particularly applicable in the 'multivariate' data collection
mode. The specific methods I will discuss are: Bootstrap Sampling and
Bootstrap Estimation. Permutation or Randomization Testing is another
computer intense method. These are also known as 'resampling' techniques.
I will demonstrate how easily these can be implemented using existing
spreadsheet software.
Handouts:
Biography: Dr. Heltshe is Professor of Statistics at the University
of Rhode Island. In addition he has been a consultant to the Narragansett
laboratory of the USEPA for the past 26 years. In this capacity he has
gained invaluable experience and exposure to 'non-traditional' data! His
research interests revolve around statistical properties of indices and
evaluation of computer intense procedures as applied to environmental
problems
|
