Online Program Home
My Program

Abstract Details

Activity Number: 540 - SPEED: Clinical Trial Design, Longitudinal Analysis, and Other Topics in Biopharmaceutical Statistics
Type: Contributed
Date/Time: Wednesday, August 1, 2018 : 11:35 AM to 12:20 PM
Sponsor: Biopharmaceutical Section
Abstract #332934
Title: Reducing the Effects of Misclassification in Sequential Multiple Assignment Randomized Trials (SMART)
Author(s): Jun He* and Donna McClish and Roy T Sabo
Companies: Virginia Commonwealth University and Virginia Commonwealth University and Virginia Commonwealth University
Keywords: SMART; Sequential Multiple Assignment Randomized Trial; Clinicial trial; Misclassification; Mixture Model; K neearest neighbor

SMART designs tailor individual treatment by re-randomizing patients to subsequent therapies based on their response to initial treatment. However, misclassification of patient responses could lead to inappropriate treatment assignment and also affect statistical analysis. To mitigate these adverse effects, patient observations can be weighted by the likelihood that their response was correctly classified. We will investigate both a mixture-model strategy and k nearest neighbor (KNN) strategy to reduce mean bias, variance bias, and improve inference at final stage outcome. The mixture model estimates the early stage probabilities of being a responder for each patient through optimizing the likelihood function by EM algorithm, while KNN estimates these probabilities based upon classifications for the k nearest observations. We expect this process fits nicely within the SMART design framework since an early outcome is used for classification and a later outcome is used for analysis. Simulations are used to compare the benefits achieved by these approaches. We also make recommendations about when these methods should be used.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program