Poster C81, Sunday, March 25, 1:00-3:00 pm, Exhibit Hall C
A Novel Information Network Flow Approach for Measuring Functional Connectivity and Predicting Behavior
Sreejan Kumar1, Kwangsun Yoo1, Monica D. Rosenberg1, Marvin M. Chun1; 1Yale University
Connectome-based predictive modeling (CPM) was recently developed to predict individual differences in behavior from functional brain connectivity (FC) (Finn et al., 2015; Nature Neurosci; Rosenberg et al., 2016, Nature Neurosci). In these models, FC was operationalized as the Pearson’s correlation between brain regions’ BOLD time-courses. However, Pearson’s correlation is limited since it only captures linear relationships. We developed a more generalized metric of FC based on information flow. This measure represents FC by abstracting the brain as a flow network of nodes that send bits of information to each other, where bits are quantified through an information-theory statistic called transfer entropy. With a sample of individuals performing a sustained attention task and resting during fMRI (n = 25), we trained CPMs to predict attention from FC patterns measured with information flow. Models trained on n–1 participants’ task-based patterns were applied to an unseen individual’s resting-state pattern to predict task performance. Model predictions significantly correlated with observed performance (r=0.639, p=5e-4). For further validation, we applied our model to three independent datasets that included resting-state fMRI data and a measure of attention (Attention Network Task performance [ANT; n = 41], stop-signal task performance [SST; n = 72], or clinician-rated ADHD symptom scores [n = 113]). Our model significantly predicted individual differences in ANT and SST performance (r=-.31, p=.049; r=.34, p=.004). Thus, information flow may be a useful alternative to Pearson’s correlation as a measure of FC due to its significant theoretical foundation and success in predicting individual differences in behavior.
Topic Area: METHODS: Neuroimaging