The long term objective of this proposal is to develop statistical methodology for the analysis of nonlinear growth data. Growth data are repeated measurements over time of some characteristic of an individual. They are used in epidemiological treatment and prevention research to study individual's changing response over time, and the effects of treatment or exposures on this response. Specifically, we will (1) develop a parametric family of tumor growth curves which describes known behaviors of tumor response to therapeutic treatment, (2) develop efficient statistical techniques for estimating and analyzing characteristics of nonlinear growth curves, and relating them to other covariates, and (3) develop and implement computer software for analysis of growth curve characteristics. We propose a two-step approach which involves (1) separately fitting each individual's growth curve and estimating the growth curve characteristics and (2) analyzing these characteristics in a generalized linear model which reflects the uncertainty due to estimation error from both steps of the procedure. We will test the tumor growth curve model and the statistical methodology on tumor growth delay data from radiobiology and tumor biology data. Future directions include extending these results to multivariate analyses and pulmonary function data.
