Smith College Applied Statistics Lecture series (2004-2005)

  1. Statistical Methods for Health Policy Data
    Dr. Sharon-Lise Normand, Professor of Health Care Policy (Biostatistics), Harvard Medical School
    October 21, 2004, 3:00pm, Burton 302 (Clark Science Center)

    Health policy research involves the examination of access, costs, and consequences of health care. The goals of the research are to provide insights regarding how to organize, finance, and deliver high quality care. In this talk, I introduce several health policy research questions and review approaches to making inference on the basis of observational data, focusing on the role of the treatment assignment mechanism. Three approaches, regression adjustment, stratification, and propensity scores, will be discussed. Methods will be illustrated to determine whether regionalization of health care services causes worse patient outcomes.

  2. How I Became a Biostatistician
    Dr. Adrienne Cupples, Professor and Chair of Biostatistics, Boston University School of Public Health
    November 15, 2004, 4:10pm, Burton 301 (Clark Science Center)

    Dr. Cupples, statistician for the Framingham Heart Study and other genetic studies will talk about her journey to becoming a biostatistician. From knowing nothing about statistics or biostatistics, Dr. Cupples began a graduate program in Statistics to pursue a masters degree. Somehow she became hooked. This lecture will describe her journey to applied biostatistics, her research experience working with the Framingham Heart Study and how she came to focus on the area of statistical genetics.

  3. Characterizing genotype-phenotype relationships with application to HIV
    Dr. Andrea Foulkes, Assistant Professor of Biostatistics and Epidemiology, University of Massachusetts/Amherst Department of Biostatistics and Epidemiology
    February 14, 2005, 4:10pm, Burton 301 (Clark Science Center)

    This talk will focus on how to integrate genetic information in the analysis of data from a cross-sectional study of HIV patients. The primary aim of this study was to assess the mediating role of genetic polymorphisms on the risk of cardiovascular complications resulting from exposure to specific antiretroviral therapies (ARTs). The use of potent ART is associated with a cluster of metabolic complications, including dyslipidemia, fat redistribution, and insulin resistance. These abnormalities are associated with increase cardiovascular risk in the general population and therefore are of particular concern for patients with HIV, for whom life-long ART may be required to maintain control of viral replication. In fact, recent studies suggest that ART, particularly protease inhibitor (PI)-based therapy, is associated with increased atherosclerotic events in HIV patients. Thus, developing strategies for identification of HIV subjects who are at increased risk of ART-related dyslipidemia will facilitate rational decision making when selecting both anti-viral and preventive cardiovascular treatment regimens.

  4. The Waterworks problem
    Dr. Herman Chernoff, Professor of Statistics, Harvard University
    March 10, 2005, 3:00pm, Burton 301 (Clark Science Center)

    A legal problem concerning the value of a waterworks system, to be taken by eminent domain, raises many issues in statistical practice. To evaluate the state of the mains carrying the water, a sample of 60 locations are taken where the pipes are exposed and assessed. How should the assessed values be presented and used? The notions of random sample, mean, standard deviation, histogram, confidence interval, simulation, experimental design, and the central limit theorem are raised. Why we should take 60 locations rather than 30 or 120 may be discussed in terms of utility theory.

  5. Integrating DNA Motif Discovery and Genome-Wide Expression Analysis
    Dr. Erin Conlon, Assistant Professor of Mathematics and Statistics, University of Massachusetts/Amherst Department of Mathematics and Statistics
    March 28th, 2005, 4:10pm, Burton 301 (Clark Science Center)

    We have designed and implemented a novel method to identify regulatory motifs in DNA sequence using the combination of microarray and DNA sequence data. We first identify a set of co-expressed genes in a microarray study and find common motifs in the regulatory sequence upstream from these genes. The candidate motifs are statistically tested for association between the motif occurrence in each gene's regulatory region and the global gene expression pattern to determine significant regulatory motifs. A multiple linear regression model identifies motifs that work in combination to control gene expression. I will illustrate our method using both single slide and time course experiments in Saccharomyces cerevisiae.

Applied Statistics Lecture series (2003-2004) Thanks to the Smith College Lecture committee, Department of Mathematics and Office of the Dean of the Faculty for support of the series.

Other 5 college seminars of interest:
University of Massachusetts Statistics and Probability Seminar Series

Organized by Nicholas Horton.
Last updated May 20, 2005