This chapter describes general issues related to modeling with the EPICURE regression modules (AMFIT, GMBO/PECAN, and PEANUTS. The modules are controlled either by commands typed by the user in the Command Line text box, executed from the Script Window, or read from command files (scripts). In addition many of the commands can be defined using menu items and dialogs (which are not described in detail here). Using the EPICURE Programs, and EPICURE Commands of the EPICURE Command Summary contain detailed examples for each program.
Since commonly used models can be specified with the FIT command, understanding all the material in this chapter is not a necessary prerequisite to the use of the regression modules.
This chapter begins with a discussion of the key variables necessary for the EPICURE modeling modules. This is followed by a description of models that can be defined and fit with the FIT command. The next section describes how categorical variables may be used in data description, model specification, and the definition of stratification variables. The next several sections of this chapter describe generalized models available in EPICURE. The topics discussed include model specification and fitting, hypothesis testing, and confidence bounds. As indicated above, it is not necessary to understand these generalized models to benefit from the use of EPICURE. However, these models are important for many risk estimation problems, especially those involving dose-response and dose-effect modification. An understanding of the nature of the generalized risk models and how to fit them will make EPICURE more useful. This chapter concludes with a discussion of subset selection and user control of various fit options.