Covariate adjustment in estimating the area under the ROC curve with partially

 
   Lead Investigator:    Danping Liu
   Institution      :    NACC
   E-Mail           :    danpingliu@gmail.com
   Proposal ID      :    180


   Publications:


1  Liu D, Zhou A
   Semiparametric estimation of the covariate-specific ROC curve in presence of ignorable verification bias
   Journal Article  Biometrics,67:906-16,2011,PMC3618959

   PUBLISHED



2  Liu D, Zhou XH
   Covariate adjustment in estimating the area under ROC curve with partially missing gold standard
   Journal Article  Biometrics,69:91-100,2013,PMC3622116

   PUBLISHED



3  Liu D,  Zhou XH
   Estimation of ROC curve with multiple types of missing gold standard
   2015   Book Section  In: Chen Z, Liu AQ, Y., Tang L, Ting N, Tsong Y, eds. Applied Statistics in Biomedicine and Clinical Trials Design. ICSA Book Series in Statistics: Sp   75-87   , 

   PUBLISHED





   Proposal Description:
   In ROC analysis, one needs to adjust for covariates if they impact the magnitude or accuracy of the test under study. Meanwhile, for many large scale screening tests, the true condition status may be subject to missingness because it may be expensive and/or invasive to ascertain the disease status. This paper addresses the issue of covariate adjustment and verification bias and proposes several estimators for the area under the covariate-specific and covariate-adjusted ROC curves (AUC$_{x}$ and AAUC). The AUC$_{x}$ is directly modelled in the form of the binary regression, and the estimating equations are based on the U statistics. The AAUC is estimated from the weighted average of AUC$_{x}$ over the covariate distribution of the diseased subjects. We employ reweighting and imputation techniques to overcome the missing data problem. Our proposed estimators are initially derived assuming that the true disease status is missing at random (MAR), and then with some modification, the estimators can be extended to the not-missing-at-random (NMAR) situation. The asymptotic variances are derived for the proposed estimators. The finite sample performance is evaluated by a series of simulation studies. Our method is applied to a data set in Alzheimer's disease research.