A0942
Title: Covariate-guided mixture of multivariate time series experts for interpretable analysis of fNIRS data
Authors: Haoyi Fu - University of Pittsburgh (United States)
Lu Tang - University of Pittsburgh (United States)
Ori Rosen - University of Texas at El Paso (United States)
Alison Hipwell - University of Pittsburgh (United States)
Theodore Huppert - University of Pittsburgh (United States)
Robert Krafty - Emory University (United States) [presenting]
Abstract: Similar to other measures of brain function, functional near-infrared spectroscopy (fNIRS) data take the form of heterogeneous multivariate time series signals. A novel group-based method for analyzing fNIRS simultaneously clusters subject-level data into potentially interpretable phenotypes is discussed while evaluating associations with clinical and demographic variables. The method models subject-level fNIRS data through a mixture of nonparametric time components where mixing weights depend on time-independent exogenous variables and account for heterogeneity among subjects. The proposed method is motivated by and illustrated through data analysis from a study of infant emotional reactivity and recovery from stress.