Latent Class Modeling

My primary research is focused on the development of statistical methods for problems in which the process of interest is unobservable. In many medical studies, the definitive outcome is inaccessible, and a valid surrogate endpoint is then measured in place of the clinically most meaningful endpoint. I have developed a latent variable model for analyzing this kind of data structure [1]. The model summarizes the unobservable definitive outcome as an underlying categorical variable and incorporates covariate effects on both underlying and measured variables. Significantly, I develop a model framework that guarantees identifiability of the two types of covariate effects [1, 3]. I also provide theory and practical methods for selecting the number of underlying variable categories [2]. The proposed approach is based on an analogous method used in factor analysis and does not require repeated model fitting under different numbers of categories.

Multiple Ordinal Measurements

Analysis of "multiply-measured" ordinal outcomes is another research topic. Co-authors and I have detailed challenges and strategies for analyzing such data [4]. We also apply generalized estimating equations methodology for analyzing multiple ordinal measurements and develop graphical diagnosis displays to evaluate the adequacy of models [5].

Joint Analysis of Transition Rates

Currently, I am working on analyzing age-related maculopathy (ARM): a leading cause of vision loss in people aged 65 or older. ARM is distinctive in that it is a disease which can progress, regress, disappear and reoccur. We develop a transitional model for jointly studying the relationship of incidence, progression, regression and disappearance rates with risk factors [6]. The developed method can be widely applied to other diseases with similar transitional characteristics.

Evaluating Diagnostic Tests While No Gold Standard

When comparing diagnostic tests of a certain disease, it is often that none of the tests can be considered as a gold standard. As a result, tests' sensitivities, specificities and Kappa statistics, which are the most commonly used indices to determine the validity and reliability of a test, cannot be obtained. Because of the ability in estimating the underlying definite outcome, latent variable model is a perfect tool for evaluating the validity and reliability of diagnostic tests while no golden standard exists. We have developed a framework for simultaneously estimating a test's sensitivity, specificity and Kappa statistic under our latent variable model [7].

Collaboration and Consultation

The Salisbury eye evaluation project, a population-based, prospective study of how vision affects functioning in older persons, attracts my attention to aging and quality of life research. Data from this project also motivates my thesis research in latent variable modeling. Investigators of the project and I have developed strategies for analyzing multiply measured self-reported visual disability in older persons [8].

The Beaver Dam Eye Study is a longitudinal population-based cohort study that aims at determining the long-term course of common vision-threatening conditions in adult Americans. I have used its data to illustrate my methodological research in the transition model and the diagnostic test evaluation. We also study the birth cohort effect in this population and propose a strategy to handle issues of longitudinal measurements and risk factor adjustment for analyzing the birth cohort effect [9].

The Wisconsin Diabetes Registry Study is a population-based cohort study that has followed individuals longitudinally from Type 1 diabetes diagnosis since 1987. In the study, we examine the association between quality of life (QOL) measurement and risk factors and propose a random effects model to draw within-individual trend of QOL [10].

I also work on several other projects. My colleagues and I have applied statistical methods for longitudinal data, missing values and functional regression to these studies.