Design, Modelling, and Computation in Studies Addressing Etiology

This thesis comprises three distinct research
papers. The first paper focuses on two important
methodological issues pertaining to the design and
analysis in a relatively new epidemiological study
design, called the case-crossover design. This
design, which only uses the cases, can be useful in
developing hypotheses regarding the etiology of an
acute event by examining the association between a
recurrent exposure and the acute event. The second
paper addresses model uncertainty in model-based low
dose extrapolation for microbial risk assessment.
Here, inference on low-dose risk estimates is highly
sensitive to model choice. We propose a new
approach called profiled Bayesian model averaging
(PBMA) to account for model uncertainty. PBMA only
requires prior distribution on the target of
inference, and can be justified based on practical
and theoretical (asymptotic) arguments. The third
paper presents simple and globally convergent
numerical methods for accelerating the convergence
of the Expectation-Maximization (EM) algorithm, a
popular approach in computational statistics
for finding maximum likelihood estimates of
parameters.

Autorentext
Dr. Ravi Varadhan is an Assistant Professor in the Division of Geriatric Medicine and Gerontology at the Johns Hopkins School of Medicine, Baltimore, USA. He has two PhDs in environmental engineering and biostatistics. His research focuses on mathematical models of the geriatric syndrome of frailty, and on computational statistics.

Klappentext
This thesis comprises three distinct research papers. The first paper focuses on two important methodological issues pertaining to the design and analysis in a relatively new epidemiological study design, called the case-crossover design. This design, which only uses the cases, can be useful in developing hypotheses regarding the etiology of an acute event by examining the association between a recurrent exposure and the acute event. The second paper addresses model uncertainty in model-based low dose extrapolation for microbial risk assessment. Here, inference on low-dose risk estimates is highly sensitive to model choice. We propose a new approach called profiled Bayesian model averaging (PBMA) to account for model uncertainty. PBMA only requires prior distribution on the target of inference, and can be justified based on practical and theoretical (asymptotic) arguments. The third paper presents simple and globally convergent numerical methods for accelerating the convergence of the Expectation-Maximization (EM) algorithm, a popular approach in computational statistics for finding maximum likelihood estimates of parameters.