Approximate Estimates of a New Class of Semiparametric Nonlinear Mixed Effect Models

摘要

Semiparametric reproductive dispersion nonlinear mixed effect model (SRDNMM) is an extension of semiparametric nonlinear mixed effect models and nonlinear reproductive dispersion models. It appears to be very flexible and useful in the analysis of complex longitudinal data not only because the covariates are often introduced into explain the inter-individual and intra-individual variations as well as an unknown time effect by a nonlinear way, but also because it allows for a variety of distributions which is much wider than exponential family for response. Although for semiparametric reproductive dispersion nonlinear mixed effect model, the likelihood method is a standard approach for inference, it can be computationally challenging. As a result, it is necessary for us to develop some computationally efficient approximate methods, In this article, we propose a novel approximate estimate method for a new class of semiparametric nonlinear mixed effect models and a simulation study is used to illustrate the proposed methodologies.