The first section of output in the NONMEM report file consists of information relating to output from the NONMEM portion of the system. In Sections 6.3.1– 6.3.6 the specific types of output in the NONMEM report file will be reviewed in additional detail.
The NONMEM report file is a text file consisting of output from the various components of the NONMEM system relevant to a particular model fitting, in addition to error messages and other useful diagnostic data. Many of the additional files, now produced with version 7, such as the *.phi and the *.ext files, can be very useful to the user interested in programmatically parsing these output files for certain pieces of data to use as an input into another helper program or application. This file is the primary repository of results for a given model. While each of these files contains important data and diagnostic information about a particular run, this chapter will focus primarily on reviewing the contents of the NONMEM report file. These include the *.rpt file, the *.FDATA file, the *.PRDERR file (possibly), the *.FCON file, and any user-requested files, such as the output from the $TABLE step(s) and/or a model specification file, and, with NONMEM version 7, the *.phi file, the *.ext file, the *.coi file, the *.cor file, and the *.cov file (Beal et al. The output described in this chapter will specifically focus on that of the classical estimation methods in NONMEM, though many of the comments are relevant to all of the estimation methods.Īfter running a given typical model with NM-TRAN and NONMEM, several different files are available to the user. Synthesizing this information to gain a complete and accurate picture of what can be learned from this particular model fit is essential to taking a logical path to a final model. Although each modeling exercise is somewhat unique in its particular goals and objectives, the NONMEM output files contain a vast amount of information addressing various aspects of each model. With enough time and experience comes a knack for zeroing in on the most relevant bits of information from among the plethora of data provided in the output files in order to make appropriate decisions about what steps to take next. The ability to understand the information provided in the various output files and to make appropriate decisions based on this information is one of the more challenging aspects of population modeling. The appropriate interpretation of the output from NONMEM may well be considered the art in what is often referred to as the art and science of population modeling.
0 kommentar(er)
