NLMEM - A NEW SAS IML MACRO FOR HIERARCHICAL NONLINEAR MODELS/

Analysis of longitudinal data is one of the most challenging tasks in statistical modeling. In the analysis, it is often necessary to take i nto account nonlinear response to a set of parameters of interest and correlation between measurements taken from the same individual. In ad dition, between-and within-subject variation has to be handled properl y. An example of addressing these issues is the hierarchical nonlinear model, where parameter estimation can be performed using linearizatio n method. In this paper a new NLMEM SAS/IML macro for hierarchical non linear models is proposed. The program uses a portion of the code deve loped earlier in NLINMIX. NLMEM retains all the benefits of NLINMIX wh ile allowing the systematic part of the model structure to be specifie d using IML syntax. Consequently, NLMEM allows estimation of models wh ich are not tractable using NLINMIX. In particular, it allows us to ad dress advanced population pharmacokinetics and pharmacodynamics models specified by ordinary differential equations. (C) 1998 Elsevier Scien ce Ireland Ltd. All rights reserved.