HACD Biostatistics is one of several research
elements comprising the Human Adaptation and Countermeasures
Division (HACD) at the Johnson Space Center. This laboratory
provides statistical consulting to HACD and the Space Medicine
Health Care Systems Office (SMHCSO), provides opportunities
for high school and college students to be directly involved
in the analysis and interpretation of biomedical research
at NASA, and conducts independent research to address the
special challenges raised by the idiosyncrasies of data often
gathered on small numbers of human subjects under non-standard
environments and test regimens.
Statistical Consulting
Biostatistics provides consulting
expertise, mainly to the HACD research laboratories,
in the application
of statistical theory and practice to ongoing biomedical
research. Laboratory personnel often aid in the preparation
of parts of research proposals that describe the experimental
design, statistical modeling and subsequent analysis
of anticipated research data. Once data is gathered,
BSL statisticians
also can assist with analysis and interpretation of results
to help the investigators extract the most information
consistent with the goal of maintaining statistical integrity.
A BSL statistician may in fact be a co-investigator in
projects requiring sophisticated statistical modeling
and/or analysis techniques
and will be expected to contribute descriptions of these
techniques in forthcoming research
papers. In these instances, the participating BSL statistician
would be included as a co-author of such papers. Being
involved as a consultant to other Bioastronautics research
laboratories
provides an excellent opportunity for the BSL statistician
to expand his/her knowledge base in such diverse medical
fields as environmental physiology, osteopathy, neurology,
pharmacology, microbiology, cardiology, nutrition and
psychology. Although HACD research laboratories are the
laboratory's
main customer, consulting support is also provided to
the SMHCSO in support of NASA flight operations.
Outreach
Although the primary customers for the BSL reside within
the Office of Bioastronautics, statistical consulting support
is occasionally given to other organizations within the
Johnson Space Center, such as the Engineering Directorate
and Human Resources and Education Office. The BSL also
provides a venue under which high school or college students,
as summer interns, can be directly involved in the analysis
and interpretation of NASA biomedical research data. Students
assigned to the BSL have a rare opportunity to become exposed
to research in a variety of biomedical fields of research,
thus facilitating their choice of career in later life.
Statistical Research
In addition to consulting duties described
above, BSL personnel are encouraged to enhance their professional
careers by
carrying out independent research to address the special
challenges raised by the idiosyncrasies of data often
gathered on small numbers of human subjects under non-standard
environments
and test regimens. Most Bioastronautics experiments aim
to certify proposed countermeasures as effective against
adverse effects of space flight on the human physiology.
Often, these experiments involve repeated observations
on disparate subjects with discrete or otherwise non-normally
distributed response variables. Proposed dependent measures
of countermeasure effectiveness may be innovative, with
little or no available published information on their variability
or co-variability, especially under proposed test conditions.
As a result, simple power calculations based on a comparison
of means, such as a t-test do not realistically quantify
design efficiency. For effective analysis, statistical
models and methodology
often have to be developed in a customized manner for
each research project. One promising area of research
would
be to investigate the feasibility of using Bayesian models
with Markov Chain Monte Carlo for accommodating the many
various random components present in Bioastronautics
research data. Another major statistical challenge is
to reduce
bias caused by severe constraints on human subject participation.
Ideally, human subjects are selected with age distribution
and other physical characteristics approximately matching
the existing NASA astronaut corps, but this is not always
possible. For example, subjects volunteering for bed-rest
studies are likely to be sedentary and not have life-styles
similar to those of astronauts. As a result, extension
of bed-rest test results to an astronaut population may
be unrealistic. In general, development of customized
methodology for modeling and correcting for informative
dropout would
be welcomed.
|