The complication in analyzing tumor data is that the tumors detectedin a screening program tend to be slowly progressive tumors, whichis the so-called length-biased sampling that is inherent inscreening studies. Under the assumption that all subjects have thesame tumor growth function, Ghosh (2008) developed estimationprocedures for proportional hazards model. In this article, bymodeling growth function as a function of covariates, we demonstratethat Ghosh (2008)'s approach can be extended to the case when eachsubject has a specific growth function. A simulation study isconducted to demonstrate the potential usefulness of the proposedestimators for the regression parameters in the proportional andadditive hazards model.
