STATISTICAL METHODS IN HUMAN DEVELOPMENT/CLINICAL STUDIES

点击翻译按钮获取中文摘要

Summary of Work: This project seeks to develop new statistical techniques for problems in human development and to apply existing techniques to those problems in novel ways. Work continues in four areas: (1) statistical modeling of the birth weight distribution in human populations, (2) regression modeling of developmental trajectories of reproductive hormones in children through puberty, (3) regression modeling of the effects of soy diet on cholesterol and sex hormone binding globulin, (4) developing methods to test whether log link or the logit link is better suited to a given data set. Development of a satisfactory model for birthweight distributions will facilitate a refined understanding of the relationship of birthweight to infant mortality. We are applying a model that we developed to U.S. birthweight and perinatal mortality records to assess how well mortality rates are predicted by distributional features measured by our technique. Having found that certain nonlinear regression models can parsimoniously describe changes in mean hormone levels through puberty, we are now examining what similar models can tell us about changes through puberty in the degree of regularity (assessed using entropy measures) exhibited in a series of 49 consecutive serum hormone measurements over 8 hr. The degree of regularity should increase as children mature.