Brazilian Journal of Probability and Statistics
- Braz. J. Probab. Stat.
- Volume 31, Number 4 (2017), 765-800.
Studying the effective brain connectivity using multiregression dynamic models
The Multiregression Dynamic Model (MDM) is a multivariate graphical model for a multidimensional time series that allows the estimation of time-varying effective connectivity. An MDM is a state space model where connection weights reflect the contemporaneous interactions between brain regions. Because the marginal likelihood has a closed form, model selection across a large number of potential connectivity networks is easy to perform. With application of the Integer Programming Algorithm, we can quickly find optimal models that satisfy acyclic graph constraints and, due to a factorisation of the marginal likelihood, the search over all possible directed (acyclic or cyclic) graphical structures is even faster. These methods are illustrated using recent resting-state and steady-state task fMRI data.
Braz. J. Probab. Stat., Volume 31, Number 4 (2017), 765-800.
Received: September 2016
Accepted: August 2017
First available in Project Euclid: 15 December 2017
Permanent link to this document
Digital Object Identifier
Mathematical Reviews number (MathSciNet)
Zentralblatt MATH identifier
Costa, Lilia; Nichols, Thomas; Smith, Jim Q. Studying the effective brain connectivity using multiregression dynamic models. Braz. J. Probab. Stat. 31 (2017), no. 4, 765--800. doi:10.1214/17-BJPS375. https://projecteuclid.org/euclid.bjps/1513328767