Parameter Estimation For A Linear Model With Interval Censoring

Regression analysis with covaxiant variables is important in many survival analysis problems. In this paper we consider the method of estimating parameters for a linear model with interval censoring.Notice the difference between interval censoring and the usual right censoring, we only know whether the survival time lies inside an interval rather than the exact value, this kind of data form is a common occurrence actually. Generally, we divide interval censored data into two types [18]. Many scholars have studied the different survival models with type I interval censored data, however, most of these research results regard a covaxiant variable as one that can be observed exactly. So interval-censored observations of a co-variant variable are infrequent. We have not found any literature on type I interval censored data when covaxiant variables are censored by now.In this paper we consider the model Y = θ'Z+ε , where we take Z to be linearly related to Y with the type I interval censored data..We have given the expressions of estimates about distribution of z and model parameters by use of the thought on self-consistency and maximum likelihood. Our research is also given proof of consistency of estimates. The result of the simulation demonstrates our research is effective and reasonable.