Our continuing research is based on four general interests and areas of expertise, which run through all our specific aims: (i) reference Bayesian methods, based on prior distributions chosen by some formal role, (ii) model selection and model averaging, (iii) sensitivity to modeling assumptions, and development of more flexible models, and (iv) computational techniques. Our research is aimed at grappling with a series of problems that are at once pressing in practice and fundamental. A continuing objective of our work is to advance the use of Bayesian methods in the biomedical and biobehavioral sciences, particularly in clinical trials, analysis of longitudinal studies, and diagnostic classification. The general methodological results we expect to obtain are motivated in part by our vigorous participation in several cross-disciplinary domains including cancer, mental health, and neuroscience, which we describe briefly in the Methods section. Through application of our methods will also learn about their practical value and discover any limitations they may have. Elaboration of simple parametric models has been a major theme in Statistics in the latter part of this century. One of the successful ideas has been to introduce covariates with regression-like structure into models for non-normal data; another is to specify parameters of interest, but let the remainder of the model remain non-parametric; and a third is to suppose a parameter itself follows a distribution. Our proposed research incorporates the first two ideas in several places, but is concerned primarily with the last notion, involving hierarchical models, mixture models, and latent variable models. With increased computing power, especially using Markov chain Monte Carlo (MCMC) methods, these random-parameter models have become central to much current statistical activity. Yet, despite great progress in the use of these models, fundamental issues remain, as we explain below, this is the main motivation of our work.

Agency
National Institute of Health (NIH)
Institute
National Cancer Institute (NCI)
Type
Research Project (R01)
Project #
2R01CA054852-08
Application #
2760201
Study Section
Special Emphasis Panel (ZRG7-STA (01))
Program Officer
Erickson, Burdette (BUD) W
Project Start
1991-07-03
Project End
2002-12-31
Budget Start
1999-01-19
Budget End
1999-12-31
Support Year
8
Fiscal Year
1999
Total Cost
Indirect Cost
Name
Carnegie-Mellon University
Department
Biostatistics & Other Math Sci
Type
Schools of Arts and Sciences
DUNS #
052184116
City
Pittsburgh
State
PA
Country
United States
Zip Code
15213
Seltman, Howard; Roeder, Kathryn; Devlin, B (2003) Evolutionary-based association analysis using haplotype data. Genet Epidemiol 25:48-58
Tzeng, Jung-Ying; Devlin, B; Wasserman, Larry et al. (2003) On the identification of disease mutations by the analysis of haplotype similarity and goodness of fit. Am J Hum Genet 72:891-902
Devlin, B; Roeder, Kathryn; Wasserman, Larry (2003) Analysis of multilocus models of association. Genet Epidemiol 25:36-47
Behrmann, Marlene; Ghiselli-Crippa, Thea; Sweeney, John A et al. (2002) Mechanisms underlying spatial representation revealed through studies of hemispatial neglect. J Cogn Neurosci 14:272-90
Evans, Julia L; Viele, Kert; Kass, Robert E et al. (2002) Grammatical morphology and perception of synthetic and natural speech in children with specific language impairments. J Speech Lang Hear Res 45:494-504
Daniels, M J; Kass, R E (2001) Shrinkage estimators for covariance matrices. Biometrics 57:1173-84
Lee, H K (2000) Consistency of posterior distributions for neural networks. Neural Netw 13:629-42
Olson, C R; Gettner, S N; Ventura, V et al. (2000) Neuronal activity in macaque supplementary eye field during planning of saccades in response to pattern and spatial cues. J Neurophysiol 84:1369-84
Carroll, R J; Roeder, K; Wasserman, L (1999) Flexible parametric measurement error models. Biometrics 55:44-54
Stangl, D K; Greenhouse, J B (1998) Assessing placebo response using Bayesian hierarchical survival models. Lifetime Data Anal 4:5-28

Showing the most recent 10 out of 18 publications