Study On Risk-Adjusted CUSUM Based On AFT Model With Measurement Error

The cumulative control charts are applied to medical clinical operations to provide accurate medical quality monitoring for medical researchers and doctors.During the monitoring process,most of the covariates affecting the patient's physical characteris-tics are obtained through measurement.However,measurement errors often exist in the measurement process.In the past studies,most of the researchers believe that these covariates are accurate measurements,or think that these measurement errors can be ignored.If monitoring with ignoring measurement error,there is always a deviation of monitoring results.This article focuses the influence of covariates with measurement error on the monitoring result.Firstly,the article briefly introduced the types of measurement error and measurement error model,designed the CUSUM control chart of risk adjustment,and provided control line h and calculation method of average run length.Then using the maximum likelihood,Newton iteration and SIMEX method to estimate the parameters of the accelerating failure model with the measurement error of the covariate,and com-paring the influence of the two parameters on the estimated model parameters without considering the measurement error and with the measurement error.Based on the accel-crated failure model of covariate with measurement error,the author proposed control charts with risk adjustment to monitor the patient's average survival time.Through the simulation study,the author compared the impact of considering measurement error on monitoring process with that of without considering measurement error.Finally,taking the data of the Busselton community health survey in Australia for example,the author illustrate that the monitoring result with considering the measurement error is more accurate than that without the measurement error.