Abstract #3552

Connectivity Analysis Through Structural Vector Auto-Regressive (SVAR) Modeling

Gang Chen¹, Ziad S. Saad¹, J. Paul Hamilton², Ian H. Gotlib², Robert W. Cox¹

¹SSCC/DIRP/NIMH, National Institutes of Health, Bethesda, MD, United States; ²Mood & Anxiety Disorders Laboratory, Department of Psychology, Stanford University, Stanford, CA, United States

Between the two popular methods in connectivity analysis, vector auto-regression (VAR) faces a challenging issue in data sampling rate while structural equation modeling (SEM) usually suffers from the violation of the assumption that no lagged correlation is considered within and across regions. With the synthesis of both methods, structural vector auto-regressive (SVAR) modeling accounts for both contemporaneous and delayed effects among the regions, and provides a more powerful and robust tool for network modeling than VAR and SEM when they are applied alone. Here we present an SVAR program that is platform-independent and open source.