Generalized Mixed Functional Modeling Approach for Discrete Scalar Outcomes and Account for the Cross-Dependence of Repeated Functional Observations

• Activity Number: 420 - Contributed Poster Presentations: Health Policy Statistics Section
• Type: Contributed
• Date/Time: Tuesday, July 30, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Mostafa Zahedjahromi* and Trent L Lalonde
• Companies: University of Northern Colorado and University of Northern Colorado
• Keywords: Binary longitudinal data; Functional data analysis; Mixed models; Longitudinal FDA; Intensive longitudinal data

Abstract

In the functional data analysis (FDA) literature, it is commonly assumed that the data are a sample of random functions varying smoothly over their observation domain (Ramsay and Silverman, 2005). Recently, longitudinal functional data analysis with continuous responses have received much attention in areas such as public health, addiction, and behavioral health. In this paper, we present a method for longitudinal FDA with a binary response, which we call longitudinal functional logistic regression. The proposed model is a generalized mixed functional modeling approach to fit a class of model for discrete scalar outcomes and account for the cross-dependence of repeated functional observations. As an application, we present analyses of intensive functional longitudinal data collected via accelerometers worn by elementary children with the purpose of assessing associations between daily activity patterns and academic performance. We show that there is an association between activity patterns and the likelihood of proficient performance on the DIBELS and AIMS standardized assessments.