Conference Program

A Bayesian Framework for Measurement Error Correction with Repeated PRO Measures (303665)

View Presentation

Chul Ahn, FDA 
Saryet Kucukemiroglu, FDA/CDRH 
Xuefeng Li, US Food and Drug Administration 
Manasi S Sheth, US Food and Drug Administration 
*Bin Wang, US Food and Drug Administration 

Keywords: Measurement error, non-parametric, patient-reported outcome

Measurement errors exist in patient-reported outcomes (PROs) and can play a significant role in effectiveness evaluations. Collecting repeated measures make it possible to model the within-subject variation and measurement errors, and hence to improve the efficiency of statistical data analysis. A Bayesian hierarchical model has been developed to adjust the impacts of the measurement errors in effectiveness evaluations based on PRO data with repeated measures. In addition, a more general approach has also been proposed to estimate the treatment effects using the deconvoluting kernel density estimator.