PDE Seminar
Center for Statistical Sciences Seminar
Abstract: In this talk we review some methods for meta-analysis of diagnostic test accuracy data, put into the framework of the (generalized) linear mixed model. We start with meta-analysis of sensitivities or specificities separately. We compare the use of the usual approximate normal within study likelihood with the exact binomial within study likelihood. Then we consider meta-analysis of the diagnostic log odds ratio and advocate the use of the non-central hypergeometric within study likelihood instead of the approximate normal as in the standard approach. Next we consider the bivariate approach to meta-analysis where per study one pair of estimated sensitivity and specificity is available. Again we advocate the routine use of the binomial within study likelihood. At least 5 different reasonable definitions of summary ROC curves could be based upon the estimated bivariate normal distribution, among them the ones proposed by Littenberg & Moses (1993) and Rutter & Gatsonis (2001). We will argue that estimation of the "true" summary ROC rests upon an untestable assumption on the way the reported sensitivity and specificity pairs were selected in the individual studies included in the meta-analysis. This makes the interpretation of the many summary ROC curves reported in the literature very problematic. Finally we discuss the extension of the bivariate model to the situation where a single diagnostic test is administered and the results are reported in a fixed number of ordered categories.
Brown University Center for Statistical Sciences Seminar
Abstract: Evaluation of diagnostic test accuracy and observer agreement have been two active areas of statistical research. Here we recognize an important link between those two areas, especially for ordinal diagnostic test/gold standard and observer agreement for ordinal measurements. For ordinal diagnostic test and gold standard test, we propose two conditional probability based indices, right agreement fraction (RAF) and left agreement fraction (LAF) who are the equivalence of sensitivity and specificity for binary diagnostic test. We argue RAF and LAF offer a more complete assessment by utilizing the multi-dimensions of the data. Next, we define two quantities derived from RAF and LAF, agreement ratio (AR) and spread (SP), which together can be used to quantify the observer agreement. We show AR and SP can be extended to evaluate continuous measurement as well. Finally, regression method applied to AR and SP can be used to compare the agreement between multiple methods or multiple raters. The definition, estimation procedure, small sample properties, as well as real data application will be included in the talk. Candidate for Assistant Professor (Research) in the Biostatistics Section of the Program in Public Health
Center for Computational Molecular Biology Seminar
Abstract: The question how genetic variation and personal health are linked is one of the compelling puzzles facing scientists today. The ultimate goal is to exploit human variability to find genetic causes for multi-factorial diseases such as cancer and coronary heart disease. Recent technology improvement enables the typing of millions of single nucleotide polymorphisms (SNPs) for a large number of individuals. Consequently, there is a great need for efficient and accurate computational tools for rigorous and powerful analysis of these data. In my talk I am going to concentrate on two computational problems, which are an essential step in studying the data obtained by this technology: Accurate and efficient significance testing with a correction for population stratification and estimating local ancestries in admixed populations.
Host: Charles Lawrence
PDE Seminar
<--- 2008 Index