Stem cell transplant is often considered the last hope for the survival for many cancer patients. The CD34+ cell content of a collection of stem cells has appeared as the most reliable indicator of the quantity of desired cells in a peripheral blood stem cell harvest and is used as a surrogate measure of the sample quality. Factors predicting the yield of CD34+ cells in a collection are not yet fully understood. Throughout the literature, there has been conflicting evidence with regards to age, gender, disease status, and prior radiation. In addition to the factors that have already been explored, we are interested in finding a cancer-chemotherapy interaction and to develop a predictive model to better identify which patients will be good candidates for this procedure. Because the amount of CD34+ cells is highly skewed, most traditional statistical methods are inappropriate without some transformation. A Bayesian generalized regression model was used to explain the variation of CD34+ collected from the sample by the cancer chemotherapy interaction. Missing data was modeled as unknown parameters to include the entire data set in the analysis. Posterior estimates are obtained using Markov chain methods. Posterior distributions identified weight and gender as well as some cancer-chemotherapy interactions as significant factors. Predictive posterior distributions can be used to identify which patients are good candidates for this procedure.

The linear mixed model is widely implemented in the analysis of longitudinal data. Inference techniques and information criteria are available and well-studied for goodness-of-fit within the linear mixed model setting. Predictive approaches such as R-squared, PRESS, and CCC are available for the linear mixed model but require more research (Edward, 2005). This project used simulation to investigate the performance of R-squared, PRESS, CCC, Pseudo F-test and information criterion for goodness-of-fit within the linear mixed model framework. Marginal and conditional approaches for these predictive statistics were studied under different variance-covariance structures. For compound symmetry structure, the success rates for all 17 statistics (marginal and conditional R-squared, PRESS, CCC, F test, AIC and BIC) were high. The study suggested using marginal rather than conditional residuals for PRESS, CCC and R-squared. It suggested using REML likelihood function which has the determinant term for AIC and BIC. For CCC, R-squared, and the information criterion, there was no difference for the various parameter number adjustments. For autoregressive order 1 plus random effect, the study suggested using conditional residuals for PRESS, marginal residuals for CCC and R-squared, and using REML function with the determinant term for AIC and BIC. Also there was no difference for the different parameter number adjustments. The F-test performed well for all covariance structures. The study also indicated that characteristics of the data, such as the covariance structure, parameter values, and sample size, can greatly impact performance of various statistics. No one criterion is consistently better than the others in terms of selection performance in the simulation study.

Chemotherapy is a method used to treat cancer but it has a number of side-effects. Research conducted by the Department of Chemical Engineering at BYU involves a new method of administering chemotherapy using ultrasound waves and water-soluble capsules. The goal is to reduce the side-effects by localizing the delivery of the medication. As part of this research, a two-factor experiment was conducted on rats to test if the water-soluble capsules and ultrasound waves by themselves have an effect on tumor growth or patient survival. Our project emphasizes the usage of Bayesian Hierarchical Models and Win-BUGS to jointly model the survival data and the longitudinal data—mass. The results of the joint analysis indicate that the use of ultrasound and water-soluble microcapsules have no negative effect on survival. In fact, there appears to be a positive effect on the survival since the rats in the ultrasound-capsule group had higher survival rates than the rats in other treatment groups. From these results, it does appear that the new technology involving ultrasound waves and microcapsules is a promising way to reduce the side-effects of chemotherapy. It is strongly advocated that the formulation of a joint model for any longitudinal and survival data be performed. For future work for the ultrasound-microcapsule data it is recommended that joint modeling of the mass, tumor volume, and survival data be conducted to obtain additional information.

Recent technological advances in fields like medicine and genomics have produced high-dimensional data sets and a challenge to correctly interpret experimental results. The Optimal Discovery Procedure (ODP) (Storey 2005) builds on the framework of Neyman-Pearson hypothesis testing to optimally test thousands of hypotheses simultaneously. The method relies on the assumption of normally distributed data; however, many applications of this method will violate this assumption. This thesis investigates the sensitivity of this method to detection of significant but nonnormal data. Overall, estimation of the ODP with the method described in this thesis is satisfactory, except when the nonnormal alternative distribution has high variance and expectation only one standard deviation away from the null distribution.

Spline smoothers as implemented in common mixed model software provide a familiar framework for estimating semi-parametric and non-parametric models. Following a review of literature on splines and mixed models, details for implementing mixed model splines are presented. The examples use an experiment in the health sciences to demonstrate how to use mixed models to generate the smoothers. The first example takes a simple one-group case, while the second example fits an expanded model using three groups simultaneously. The second example also demonstrates how to fit confidence bands to the three-group model. The examples use mixed model software as implemented in lme() in R. Following the examples a discussion of the method is presented.

This paper develops a Bayesian ordinal regression model for the maladaptive subscales of the Inventory for Client and Agency Planning (ICAP). Because the maladaptive behavior section of the ICAP contains ordinal data, current analysis strategies combine all the subscales into three indices, making the data more interval in nature. Regular MANOVA tools are subsequently used to create a regression model for these indices. This paper uses ordinal regression to analyze each original scale separately. The sample consists of applicants for aid from Utah's Division of Services for Persons with Disabilities. Each applicant fills out the Scales of Independent Behavior"”Revised (SIB-R) portion of the ICAP that measures eight different maladaptive behaviors. This project models the frequency and severity of each of these eight problem behaviors with separate ordinal regression models. Gender, ethnicity, primary disability, and mental retardation are used as explanatory variables to calculate the odds ratios for a higher maladaptive behavior score in each model. This type of analysis provides a useful tool to any researcher using the ICAP to measure maladaptive behavior.

Because of the huge amounts of data made available by the technology boom in the late twentieth century, new methods are required to turn data into usable information. Much of this data is categorical in nature, which makes estimation difficult in highly multivariate settings. In this thesis we review various multivariate statistical methods, discuss various statistical methods of natural language processing (NLP), and discuss a general class of models described by Erosheva (2002) called generalized mixed membership models. We then propose extensions of the information partition function (IPF) derived by Engler (2002), Oliphant (2003), and Tolley (2006) that will allow modeling of discrete, highly multivariate data in linear models. We report results of the modified IPF model on the World Health Organization's Survey on Global Aging (SAGE).

Pediatric injury is both common and expensive. Finding ways to prevent pediatric injury is a major public health concern. Many studies have investigated various aspects of pediatric injury, and some suggest that birth order may be an important risk factor for pediatric injury. This study further examined the relationship of birth order with pediatric injury, specifically studying the association of birth order with emergency department-attended infant injury while adjusting for other important family and individual covariates. Data for analysis included Utah birth certificate, death certificate, and hospital emergency department datasets, which were probabilistically linked to obtain complete demographic and injury information for infants born in 1999—2002. Three groups of risk factors were defined for analysis: maternal demographics, maternal risk behaviors, and infant demographics. Two outcome variables were defined for this study, “injury event” and “severe injury event.” Data was analyzed using generalized estimating equations (GEE). Birth order was associated with infant injury events and severe infant injury events. Birth order 4th or greater had the greatest effect for both injury outcomes. Additionally, several maternal characteristics were associated with infant injury events and severe infant injury events. In particular, maternal age and maternal smoking behavior were associated with increased infant injury risk. This study identified two targeted populations that are well-suited to injury prevention efforts: infants born to mothers who smoke, and infants born to mothers who are young and have many other children.

Pre-election polls are used to test the political landscape and predict election results. The relative weights for the state-level data from the 2006 U.S. senatorial races are considered based on the date on which the polls were conducted. Long- and short-memory weight functions are developed to specify the relative value of historical polling data. An optimal weight function is estimated by minimizing the discrepancy function between estimates from weighted polls and the election outcomes.

Traditional clinical trials have been inefficient in their methods of dose finding and dose allocation. In this paper a four-parameter logistic equation is used to model the outcome of Bernoulli-response clinical trials. A Bayesian adaptive design is used to fit the logistic equation to the dose-response curve of Phase II and Phase III clinical trials. Because of inherent restrictions in the logistic model, symmetric candidate densities cannot be used, thereby creating asymmetric jumping rules inside the Markov chain Monte Carlo algorithm. An order restricted Metropolis-Hastings algorithm is implemented to account for these limitations. Modeling clinical trials in a Bayesian framework allows the experiment to be adaptive. In this adaptive design batches of subjects are assigned to doses based on the posterior probability of success for each dose, thereby increasing the probability of receiving advantageous doses. Good posterior fitting is demonstrated for typical dose-response curves and the Bayesian design is shown to properly stop drug trials for clinical futility or clinical success. In this paper we demonstrate that an adaptive Bayesian approach to dose-response studies increases both the statistical and medicinal effectiveness of clinical research.