Lefschetz Center for Dynamical Systems Seminar
Brown University Center for Statistical Sciences Seminar
Abstract: We propose a new class of models, frailty measurement error models (FMeMs), for clustered survival data when covariates are measured with error. We explore FMeMs from three directions: bias analysis, structural modeling and functional modeling. We study the asymptotic bias when measurement error is ignored and when the underlying distribution of the unobserved error-prone covariates is misspecified. Structural modeling and functional modeling is developed to make statistical inference in FMeMs. Under structural modeling, we assume a distribution for the unobserved error-prone covariates and calculate nonparametric maximum likelihood estimates (NPMLEs) using an EM algorithm. Under functional modeling, we make no distributional assumption on the unobserved error-prone covariates and use the SIMEX method for parameter estimation. The NPMLEs and SIMEX estimates are compared in terms of efficiency and robustness. The proposed methods are applied to the west Kenya parasitemia data and their performance is evaluated through simulations.
PDE Seminar
Brown University Center for Statistical Sciences Seminar
Abstract: The ideal multiple reader diagnostic imaging study provides images of each case using all of the diagnostic modalities being compared, suspicion ratings for each image form all of the radiologists participating in the study, and definitive knowledge of the true disease status for each subject. In practice, such complete data are often not available. This talk presents methods for the analysis of multiple correlated ROC curves when some data are missing. The methods use ordinal regression models in conjunction with generalized estimating equations. First, we provide methodology to accommodate arbitrary patterns of missingness in the suspicion ratings when this missingness is independent of the unobserved suspicion ratings. This method can be applied in several common situations, such as when claustrophobic patients do not undergo magnetic resonance imaging. Second, we provide methodology to correct for possible verification bias. This method is useful when the gold standard is not obtained for some patients, and the decision of whether to obtain it depends on observed data. An example of this is when surgical biopsies are not obtained I patients with advanced disease who choose to receive palliative car. The statistical methods are illustrated with the analysis of data from a study in diagnostic oncology.
Department of Mathematics Colloquium
<--- 1999 Index