American Statistical Association
Advances in modern technology have facilitated the collection of data that are measured over a number of randomly distributed points (e.g. time) on the same subject, or experimental unit. Such data, often referred to as "functional data", arise commonly in health sciences research. We describe a new modeling framework for a continuous response outcome having a multilevel functional structure, where the functions at the lowest hierarchy level are spatially correlated. Our approach provides fast and robust inferential tools with particular interest in modeling the spatial correlation. The idea is to partition the total covariance of the observed model into simple functional mixed effects components, using multilevel principal components. This parsimonious orthonormal decomposition of the functional spaces leads to major computational improvements over current methods (Morris & Carroll, 2006, Baladandayuthapani et al., 2007). The proposed procedure is illustrated through simulation studies and the data from the colon carcinogenesis experimental study.
Ana-Maria Staicu received her Ph.D. in Statistics from the University of Toronto in 2007, under the direction of Nancy Reid. She did her postdoctoral studies at the University of Bristol, UK and recently joined the Department of Statistics at NC State University. Her research interests include functional data analysis, nonparametric regression, higher-order likelihood inference and conditional inference.
|Date:||Thursday, April 22, 2010|
|Time:||4:00 - 5:00 P.M.|
Mailman School of Public Health
Department of Biostatistics
722 West 168th Street
Biostatistics Computer Lab
6th Floor - Room 656
New York, New York