Some estimation problems in carcinogenesis risk assessment.

摘要

Some of the estimation problems encountered in carcinogenesis risk assessment are discussed. The discussion includes a brief introduction to risk assessment and the statistical models and inferential techniques that are used in this area. The problem of low-dose extrapolation is discussed in some detail with emphasis on low-dose linear extrapolation. Three modifications to a model-free procedure for low-dose linear extrapolation are discussed and a new procedure is introduced. An exploratory investigation is conducted into the potential use of optimal designs in improving inference in low-dose extrapolation. The primary objective is to find some means through which inference associated with low-dose extrapolation can be made less conservative. Possible parameterizations of a model that incorporates time-dependent exposure and a statistical test related to this model are presented. Further research under consideration related to the problems discussed is also highlighted. The central problem is to enhance statistical inference in carcinogenesis risk assessment. Throughout, emphasis is on practical applications and considerations.