2020/01/22 by Adam Ciarleglio, Ciarleglio, Adam, Eva Petkova +3
Agricultural and Biological Sciences · Neuroscience · #Applications (stat.AP) #FOS: Computer and information sciences #Functional Brain Connectivity Studies #Neural and Behavioral Psychology Studies #Sensory Analysis and Statistical Methods
paper · pdf · doi:10.48550/arxiv.2001.08175
openalex publication_date 2020/01/22 · openalex created_date 2022/10/02 · openalex updated_date 2026/07/28
Current source density (CSD) power asymmetry, a measure derived from\nelectroencephalography (EEG), is a potential biomarker for major depressive\ndisorder (MDD). Though this measure is functional in nature (defined on the\nfrequency domain), it is typically reduced to a scalar value prior to analysis,\npossibly obscuring the relationship between brain function and MDD. To overcome\nthis issue, we sought to fit a functional regression model to estimate the\nassociation between CSD power asymmetry and MDD diagnostic status, adjusting\nfor age, sex, cognitive ability, and handedness using data from a large\nclinical study. Unfortunately, nearly 40 % of the observations were missing\neither their functional EEG data, their cognitive ability score, or both. In\norder to take advantage of all of the available data, we propose an extension\nto multiple imputation by chained equations that handles both scalar and\nfunctional data. We also propose an extension to Rubin's Rules for pooling\nestimates from the multiply imputed data sets in order to conduct valid\ninference. We investigate the performance of the proposed extensions in a\nsimulation study and apply them to our clinical study data. Our analysis\nreveals that the association between CSD power asymmetry and diagnostic status\ndepends on both age and sex.\n