This is a continuing proposal to address a variety of biostatistical problems motivated by current issues in imaging neuroscience, as during the previous funding cycle.
New aims : the development of flexible semiparametric growth curve models for accelerated longitudinal designs; advancing methodology for replicated spatial point processes and 3-D brain brain cell assemblies; and new methods and algorithms for semiautomatic identification of human brain cells. We propose to generalize our proposed individual low-rank smooth regression methods to compositional data via a Iogit-Gaussian model within a hierarchical Bayes framework. We seek to produce practical guidelines for designing cost-effective longitudinal studies involving expensive outcomes measurements. We propose to advance Poisson random field methods for sparse processes, motivated by the multiple cell types and regional structures in the human brain. Empirical data analysis will continue to play a central role in the proposed research. Our human brain mapping research by magnetic resonance imaging (MRI) and positron emission tomography (PET) and human brain tissue microscopy again relates directly to the study of psychiatric and neurological outcomes in healthy and ill subjects, both young and old. Through our collaborating biostatistical and neuroscience institutions, our ongoing translational research develops and links modem biostatistical methods with complementary work in longitudinal anatomic human brain imaging, functional human brain imaging and human brain tissue microscopy. Brain diseases addressed are schizophrenia, bipolar disorder and Parkinson's disease. However, potential applications of our methods go well beyond human brain mapping to include longitudinal and spatial epidemiology, risk assessment, health policy and management, nutrition, and other fields in which cost and feasibility constraints impose restrictions on the numbers of subjects studied and on the numbers and timings of their repeated measurements.
Showing the most recent 10 out of 24 publications