Inference in high dimensional linear measurement error models

For a high dimensional linear model with finite number of covariates measured errors, we study statistical inference on the parameters associated error-prone covariates, and propose new corrected decorrelated score test corresponding type estimator. This work was motivated by real data example, where both low phenotypic variables genotypic variables, single nucleotide polymorphisms (SNPs), are available. One is clinical interest but error. As standard in literature, SNPs assumed to be accurately. We show that limiting distribution our statistic normal under null hypothesis retains power local alternatives around zero. also establish asymptotic normality newly proposed estimator, hence confidence intervals can constructed. The finite-sample performance procedure examined through simulation studies. further illustrate via an empirical analysis example mentioned above.