Bayesian Methods for a Longitudinal CAT

University Of Tx Md Anderson Can Ctr, Houston TX

Investigators

Linked publications & trials

Abstract

[unreadable] DESCRIPTION (provided by applicant): My goal is to be a quantitative methodologist focusing in behavioral science applications. Specifically I am interested in using computerized adaptive testing (CAT) to improve longitudinal measurement of patient reported health outcomes important in cancer prevention and cancer control, such as depressive symptoms. In order to get precise measurement of patient-reported health outcomes, self-report surveys tend to be long and create a moderate-to-high level of respondent burden. CAT is a more efficient measurement tool than most current scales and reduces response burden while simultaneously increasing precision. I propose to improve longitudinal assessment by first improving CAT methodology to achieve even higher gains, and then to develop CAT measurement instruments for longitudinal measurement of patient-reported health outcomes. [unreadable] [unreadable] This career development award will build upon my previous training in statistical science and give me further training in cutting edge Bayesian statistical methods and patient reported health outcomes research. I have assembled a mentoring team of two statisticians, a psychometrician, a quality of life researcher, and one of the leaders in developing computerized adaptive tests (CAT) for measuring patient reported health outcomes. [unreadable] [unreadable] My research proposal focuses on developing methodological enhancements to current CAT algorithms, and developing new algorithms to further reduce respondent burden in longitudinal assessments of patient reported health outcomes in behavioral cancer prevention and cancer control studies. I hypothesize that tailoring CAT algorithms to longitudinal assessment will improve precision and reduce patient burden relative to current methods. I will use simulation studies to 1) compare three different methods for developing a CAT algorithm, and 2) to develop and compare CAT algorithms tailored to longitudinal assessment. The research and training in this career development award will prepare me to be a principal investigator on research projects and grants that continue methodological enhancements in CAT used to measure patient reported health outcomes in behavioral cancer prevention and control studies, and to bring this tool into the clinic for research purposes. The methodology developed in this study will be applicable to a wide range of computer hardware devices, such as desktops, laptops, and even handheld computers. [unreadable] [unreadable] [unreadable] [unreadable]