A Comparison of Methods for Imputing Missing Longitudinal fMRI Data

• Activity Number: 174
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Maria Josefsson* and Anders Lundquist and Lars Nyberg
• Companies: Umeå University and Umeå University and Umeå University
• Keywords: fMRI ; longitudinal data ; missing data ; imputing ; ageing

Abstract

Missing information is one of the main methodological problems in longitudinal studies, including longitudinal fMRI (L-fMRI) studies. One drawback with available methods for analyzing L-fMRI data is the exclusion of subjects with missing information, which may result in biased estimates and loss of power. One approach to this problem is to impute the missing information to obtain a completed dataset. Our goal with this study was to adapt available imputation methods to a setting with L-fMRI data, where some subjects had missing observations at follow-up. We applied these imputation methods to a real dataset, and further evaluated the methods in a simulation study.