Show/Hide University Links

UTSA College of Business
Management Science and Statistics News
Archives

Seminar Series: Fall 2011
The seminar series will usually take place on Fridays in a Business Building room, but the exact time and location could be different due to a variety of factors including room availability in the Business Building. For actual direction to the Business Building, please see campus map. For additional information contact Dr. Kefeng Xu, (210) 458-5388.

Friday, October 21, 2011, 2-3PM, Business Building 3.01.10

  • Presenter: Valen E. Johnson, Professor and Deputy Chair, Department of Biostatistics University of Texas MD Anderson Cancer Center, Houston, TX

  • Presentation Title: Bayesian Model Selection in High-dimensional Settings

  • Abstract: Standard assumptions incorporated into Bayesian model selection procedures result in model selection procedures that are not competitive with commonly used penalized likelihood methods. I propose modifications of these methods by imposing non-local prior densities on model parameters. I show that the resulting model selection procedures are consistent in linear model settings when the number of possible covariates p is bounded by the number of observations n, a property that has not been extended to other model selection procedures. In addition to consistently identifying the true model, the proposed procedures provide accurate estimates of the posterior probability that each identified model is correct. Through simulation studies, I demonstrate that these model selection procedures perform as well or better than commonly used penalized likelihood methods in a range of simulation settings.

  •  

Tuesday, Nov. 8, 2011, 8:00-9:30pm, Business Building 4.02.10 (Executive Conference Room), joint seminar with ASA-San Antonio Chapter

  • Presenter: Brad H. Pollock, MPH, PhD Professor and Chairman, Department of Epidemiology & Biostatistics University of Texas Health Science Center at San Antonio

  • Presentation Title: Small Area Analysis of Primary Liver Cancer Incidence in Bexar County, Texas

  • Purpose: Spatial analysis was performed examining Bexar County, Texas primary liver cancer incidence patterns for the period 1995 to 2006.
    Methods: A small area analysis was conducted using age-, sex- and race-standardized incidence rates for census tracts in Bexar County, Texas over three time periods: 1995–1998, 1999–2002, and 2003–2006. An additional analysis was performed using age-, sex- and ethnicity-standardized incidence. Both the Local Indicator of Spatial Autocorrelation (LISA) statistic and spatial scan statistic methods were used to assess geospatial clustering of the tract-level standardized cancer incidence rates. Localized rate comparisons were performed using first- and second-order spatial adjacency of census tracts for cluster regions identified from the spatial scan statistic analysis. The distribution of incidence rates for census tracts that overlapped known halogenated hydrocarbon plumes were statistically compared to non-overlapping census tracts using the Wilcoxon rank sum test.
    Results: A total of 1,403 incident liver cancer cases diagnosed between 1995 and 2006 were included in a database file supplied by the Texas Cancer Registry. Of these cases, 35 cases were excluded because of missing age, sex, or ethnicity information or because the recorded residence was outside of Bexar County, yielding 1,368 evaluable cases.
    For the geospatial analysis, with the LISA method, a total cancer cluster was found in the area between Interstate-35 and Interstate-10 on the east side of downtown San Antonio over the three time periods. However, this area did not show significant clustering when analyzed with the spatial scan statistic method. In contrast, both the LISA and spatial scan statistic methods identified a consistent liver cancer cluster in the area west of Downtown San Antonio. There was a slight shift westward and southward for this cluster over the 1995 to 2006 period. Localized rate comparisons for this liver cancer cluster showed significant differences (p<0.001) with the rest of the county using both first- and second-order spatially neighboring census tracts. We observed approximately the same cluster area with higher than expected occurrence of primary liver cancer when the age-, sex- and ethnicity-standardized rates were used instead of age-, sex- and race-standardized rates.
    For patients diagnosed with cancer in 1999–2002, the median standardized liver cancer incidence rate was 12.8/100,000 for the census tracts that overlapped the 1999–2000 halogenated hydrocarbon plume area compared to 4.6/100,000 for the tracts not that did not overlap the plume. In contrast, the median standardized total cancer incidence rate was 375.8/100,000 for census tracts that overlapped the plume area which was similar to 401.9/100,000 rate for the census tracts that did not overlap the plume area. For patients diagnosed with cancer in 2003–2006, the median liver cancer rate was 20.1/100,000 and 7.8/100,000 for the plume and non-plume tracts, respectively. Likewise, for the same patients, the median total cancer incidence rates were 358.1/100,000 and 420.1/100,000 for plume and non-plume tracts, respectively. The liver cancer rate appeared to increase over time while the total cancer rate did not. When analyzing more recent 2005 TCE plume data, the relative differences between plume and non-plume census tract incidence rates remained roughly the same as the rates using the 1999–2000 plume information.
    Conclusions: For Bexar County, there was evidence of spatial clustering of primary liver cancers in the area west of Downtown over the years 1995–2006. Clustering of total cancers was not evident. Standardized liver cancer rates appeared higher for census tracts that overlapped plume areas with detectable groundwater halogenated hydrocarbons.
    Recommendations: Recognizing the inherent limitations of ecologic analyses, ultimately, an analytic case-control study of primary liver county incidence in Bexar County would more definitively assess the impact of a number of putative risk factors, including detailed individual information on hepatitis B and C infection characteristics, chronic alcohol use, diabetes and nonalcoholic steatohepatitis (NASH), and mycotoxin exposure. 

     

Monday, Nov. 28,2011, 10am-11:30am, Stat Lab (BB 3.02.16) - Ph.D. Dissertation Defense

  • Presenter: Mr. Xiaobin Yang, Ph.D. Candidate, Dept. of Management Science and Statistics, University of Texas at San Antonio

  • Supervising Professor: Keying Ye, Ph.D.

  • Presentation Title: AN EXTENSION OF CONTINUAL REASSESSMENT METHOD IN PHASE I CLINICAL TRIALS

  • Abstract: Model-based clinical trial designs have drawn much attention from the biostatistical community since 1990 when O’Quigley et al. proposed the Continual Reassessment Method (CRM). CRM and its various modified versions have achieved great successes in finding the maximum tolerated dose (MTD) adaptively in the case of dichotomous toxicity responses (i.e. dose-limiting toxicity, DLT, or non-DLT). In dose-escalation processes, it is crucial to differentiate severity of DLT if the impact of severity of toxicity is substantial (e.g. liver toxicities). However, due to the limitation of its model structure, it is difficult to extend CRM naturally to the polychotomous toxicity responses.

    In this research, we propose a two-parameter probit model with latent variables to extend the CRM for the cases of dichotomous and polychotomouse toxicity responses. For the dichotomous toxicity responses, simulation results under different scenarios show that the proposed model is superior to the power model originally used in the CRM. We extend the proposed model naturally to the ploychotomous toxicity responses by categorizing the latent variables corresponding to the bin boundary parameters. Simulation results show that two-parameter probit model with latent variables works encouragingly well in the case of polychotomous toxicity responses. By differentiating severity of DLT, the number of patients allocated to the higher toxicity dose level is reduced. That reduces the risk of toxicity for patients in the clinical trial study.

    In addition, we introduce the concept of the overall MTD which makes it possible to study both dichotomous and polychotomous response models under a unified framework. Analytic properties of the overall MTD are given. It is shown that dichotomous response model is a special case of polychotomous response model and under certain circumstances the polychotomous response model is reduced to dichotomous responses model.

    Furthermore, we generalize the concept of the overall MTD to a broader case. Under this concept, a unified model is built, which includes categorical (such as dichotomous and polychotomous) and numerical (such as continuous) toxicity responses as its special cases. The Convergence, Robustness and Reduction theorems of the overall MTD are proved in the general case. As an example of the continuous toxicity response, the normal toxicity response is studied along with a family of the target toxicity probabilities.

 

Tuesday, November 29, 2011, 10:00-11:30am, Stat Lab (BB 3.02.16) - Ph.D. Dissertation Defense

  • Presenter: Mr. Daniel G Polhamus, Ph.D. Candidate, Dept. of Management Science and Statistics, University of Texas at San Antonio

  • Supervising Professor: Nandini Kannan, Ph.D. and Keying Ye, Ph.D.

  • Presentation Title: BAYESIAN SEQUENTIAL ANALYSIS FOR CORRELATED TIME TO EVENT DATA: A COMPUTATIONAL APPROACH

  • Abstract: The computational and analytical burden of the Bayesian sequential decision problem has historically limited its application domain. Advances in decision-theoretic loss approximation through simulation (Carlin et al., 1998, Brockwell and Kadane, 2003, Mueller et al., 2007) present an option for simpler, low-dimensional designs based upon optimization of the decision boundaries or discretization of the parameter space. The combination of these approaches and high performance computing (HPC) allow for the application of Bayesian sequential decision methodology to increasingly complex problems. For example, one such problem is the Weibull-Stable frailty model for correlated event times. The time requirement for sampling its joint posterior is non-trivial and practical simulation based approaches are infeasible without inclusion of a high level of a parallelization. With increasing public accessibility to HPC (e.g., cloud based providers such as Amazon, and Rackspace), the level of parallelization required is easily achievable. Using HPC, the utility and efficiency of the simulation based decision-theoretic methods in the context of loss functions, interim analysis, and correlated survival data is explored. This dissertation provides methodology for reaching Bayes optimal decisions for the Weibull-Stable model in a matter of hours, rather than months.

 

Thursday, Dec. 1, 2011, 8 to 9:30 pm, Business Building 4.02.10 (Executive Conference Room), joint seminar with ASA-San Antonio Chapter.

  • Presenter: Dr. Dennis Lin, University Distinguished Professor of Supply Chain and Statistics, Penn State University

  • Presentation Title: Recent Advances in Computer Experiments

  • Abstract: Computer models can describe complicated physical phenomena. However, to use these models for scientific investigation, their generally running times and mostly deterministic nature require a special designed experiments. This talk attempts to address the fundamental question of "what is a (proper) computer simulation?" The basic concepts and usefulness for various simulation issues will be discussed, no specific algorithm will be given. Second portion of the talk will be focused on design of computer experiment/simulation models. Recent advances on Latin Hypercube Design and Uniform Design will be discussed. Slides of his talk can be downloaded at the website http://www.personal.psu.edu/users/j/x/jxz203/lin/Lin_pub/

  • Brief Bio: Dr. Dennis Lin is a University Distinguished Professor of Supply Chain and Statistics at Penn State University. His research interests are quality engineering, industrial statistics, data mining and response surface. He has published over 150 papers in a wide variety of journals. Dr. Lin is an elected fellow of ASA and ASQ, an elected member of ISI, a lifetime member of ICSA, and a fellow of RSS. He is an honorary chair professor for various universities, including a Chang-Jiang Scholar of China at Renmin University, Jinan University, National Chengchi University, and Fudan University. Dr. Lin presents several distinguished lectures, including the 2010 Youden Address (FTC) and the 2011 Loutit Address (SSC). He is also the recipient of the 2004 Faculty Scholar Medal Award at Penn State University.

 

Friday, Dec. 2, 2011, 1:30 pm - 3:00 pm, Stat Lab (BB 3.02.16) - Ph.D. Dissertation Defense

  • Presenter: Mr. Liang Jing, Ph.D. Candidate, Dept. of Management Science and Statistics, University of Texas at San Antonio

  • Supervising Professor: Victor De Oliveira, Ph.D.

  • Presentation Title: BAYESIAN MODEL CHECKING FOR GENERALIZED LINEAR SPATIAL MODELS FOR COUNT DATA

  • Abstract: Hierarchical models are increasingly used in many of the earth sciences. A class of Generalized Linear Mixed Models was proposed by Diggle, Tawn and Moyeed (1998) for the analysis of spatial non-Gaussian data, but model estimation, checking and selection in this class of models remain difficult tasks due to the presence of an unobservable latent process and model checking methods have not been considered in the literature. We consider this class of models for the analysis of spatial count data. We implement robust Markov Chain Monte Carlo algorithms with the help of advanced techniques, such as group updating, Langevin-Hastings algorithms, and data-based transformations, for estimation and posterior sampling. Then we investigate the application of model checking methods based on measures of relative predictive surprise, as those described in Bayarri and Castellanos (2007). Besides, we propose and investigate an alternative model checking method to diagnose incompatibility between model and data based on a kind of transformed residuals. The usefulness of the proposed model checking methods is explored using both simulated and real spatial count data, and the results are compared with the results from other Bayesian model checking methods. An R package is also developed to implement all the methods discussed in the dissertation by using advanced computing techniques, such as R/C++ interfacing and parallel computing.

 

UTSA Homepage | College of Business Homepage | Events | Community | Giving | Directory | Contact Us | Site Map
College of Business — The University of Texas at San Antonio
One UTSA Circle — San Antonio, TX 78249-0631— (210) 458-4313 FAX (210) 458-4308

Please send comments about this site to the College of Business Webmaster
Privacy & Security Policies
TSA Homepage College of Business Homepage