On the use of covariates for rodent bioassay and screening experiments
研究了一个三状态模型，通过协变量和随机筛查使中间状态的患病率可识别且易于估计，并用啮齿动物肿瘤实验数据验证。
Consider a three-state compartmental model with an initial, an absorbing and an intermediate state which may or may not be entered before absorption. Observable data are time to absorption and an indication of whether the intermediate state was entered, but not the time of transition. The transition rates characterizing the process are nonidentifiable without simplifying assumptions or additional information on the prevalence function, defined as the time specific proportion of nonabsorbed individuals in the intermediate state. Our approach makes prevalence identifiable and easily estimable through the use of covariates, combined with one or more random screens of the population. The results are illustrated with pathology data from a rodent tumourigenicity experiment involving several random sacrifices.