The corrected likelihood approach for adjusting measurement error in Cox’s model
摘要
While dealing with survival data, we may come across situations where some of the covariate(s) affecting the failure time are prone to measurement error. Further, the effect of the model covariates on the event time may vary for different groups in the underlying population. The paper discusses about modeling and estimating the effect of mis-measured model covariates on the event times, where the effects are dependent on an auxiliary covariate that categorizes the population into groups. A corrected likelihood approach that can eliminate the bias due to the measurement error, is used to estimate the parameters.