Weibull Regression Model for Analysis of Middle-Censored Lifetime Data

Middle censoring introduced by Jammalamadaka and Vasudevan (2003) refers to data arising in situations where the exact lifetime becomes unobservable if it falls within a random censoring interval, otherwise they become observable. In the present paper we propose a proportional hazards regression model for a lifetime data subject to middle censoring, where the lifetimes are assumed to follow Weibull distribution. The regression parameters are estimated using the EM algorithm. Asymptotic normality of these estimators are established. We report simulation study for the model to assess the finite sample properties and analyse a real life data on survival times in months for Multiple Myeloma patients studied by Krall et al. (1975)