A semiparametric model for interval censored and truncated data

In this work we consider a complex observational scheme, that is, survival data that are both interval censored and truncated. We assume a semiparametric Cox model for the survival function and consider censoring and truncation distributions as in Huber, Solev and Vonta (2006, 2007). We establish the form of the least favorable model (Slud and Vonta (2005)) for the cumulative hazard function, which plays the role of the infinite-dimensional nuisance parameter, for fixed values of the finite-dimensional parameter of interest. The least favorable model cannot be defined in closed form. Bibl. – 8 titles.