Model Checking for Biomedical Data

We will develop commercial software which implements graphical and numerical methods for checking tadequacy of a variety of regression models frequently used in biomedical studies. In Phase I of theconcentrate on three popular models: the Cox proportional hazards model, the logistic regression modeffects model for longitudinal data. The model checking techniques considered here include the objecprocedures recently developed by L.J. Wei and his colleagues. In Phase II of the project, we will decommercial package. The package will consist of all the important model checking techniques in the lmodels as well as other regression models like the parametric survival model, the generalized linearextension to correlated data, and models for polytomous responses. This software will be incorporatepackage developed by Cytel Software Corporation. Both the Phase I and Phase II software will be avaiin three ways: as external procedures connected to the SAS system, as external functions in S-plus,software utility under Microsoft Windows-95.