Interface Between NONMEM Output And An Exploratory Data Analysis Tool

P. Girard, A. Laffont, JP. Boissel

Unité de Pharmacologie Clinique, Hôpital Cardiologique, Lyon

According to the defined statistical model for inter-individual variability (example CL_i = CL + η_i where i denotes the i^th patient) the new NONMEM version IV computes empirical bayesian individual η_i using FO or a conditional estimation method. These η_i can be output on an ASCII table. We will present a simple SAS macro that reads this output and computes the individual estimated parameters, in order to analyse them using an interactive exploratory data analysis (EDA) tool, named SAS/INSIGHT. Then we will demonstrate how the EDA tool, can be used to estimate empirical distribution of the individual parameters, to smooth the distribution using Rernels and to look for covariates by exploring the marginal distributions of parameters and covariates.

Reference: PAGE 2 (1993) Abstr 900 [www.page-meeting.org/?abstract=900]

Poster: oral presentation