Mapping epileptic directional brain networks using intracranial EEG data

Summary The human brain is a directional network system, in which brain regions are network nodes and the influence exerted by one region on another is a network edge. We refer to this directional information flow from one region to another as directional connectivity. Seizures arise from an epileptic directional network; abnormal neuronal activities start from a seizure onset zone and propagate via a network to otherwise healthy brain regions. As such, effective epilepsy diagnosis and treatment require accurate identification of directional connections among regions, i.e., mapping of epileptic patients’ brain networks. This article aims to understand the epileptic brain network using intracranial electroencephalographic data—recordings of epileptic patients’ brain activities in many regions. The most popular models for directional connectivity use ordinary differential equations (ODE). However, ODE models are sensitive to data noise and computationally costly. To address these issues, we propose a high-dimensional state-space multivariate autoregression (SSMAR) model for the brain’s directional connectivity. Different from standard multivariate autoregression and SSMAR models, the proposed SSMAR features a cluster structure, where the brain network consists of several clusters of densely connected brain regions. We develop an expectation–maximization algorithm to estimate the proposed model and use it to map the interregional networks of epileptic patients in different seizure stages. Our method reveals the evolution of brain networks during seizure development.

Innovations orthogonalization: a solution to the major pitfalls of EEG/MEG “leakage correction”

1.AbstractThe problem of interest here is the study of brain functional and effective connectivity based non-invasive EEG-MEG inverse solution time series. These signals generally have low spatial resolution, such that an estimated signal at any one site is an instantaneous linear mixture of the true, actual, unobserved signals across all cortical sites. False connectivity can result from analysis of these low-resolution signals. Recent efforts toward “unmixing” have been developed, under the name of “leakage correction”. One recent noteworthy approach is that by Colclough et al (2015 NeuroImage, 117:439-448), which forces the inverse solution signals to have zero cross-correlation at lag zero. One goal is to show that Colclough’s method produces false human connectomes under very broad conditions. The second major goal is to develop a new solution, that appropriately “unmixes” the inverse solution signals, based on innovations orthogonalization. The new method first fits a multivariate autoregression to the inverse solution signals, giving the mixed innovations. Second, the mixed innovations are orthogonalized. Third, the mixed and orthogonalized innovations allow the estimation of the “unmixing” matrix, which is then finally used to “unmix” the inverse solution signals. It is shown that under very broad conditions, the new method produces proper human connectomes, even when the signals are not generated by an autoregressive model.

THE ROLE OF SECTORAL SHIFTS IN THE DECLINE OF REAL GDP VOLATILITY

U.S. production has shifted from goods-producing to service-producing industries. We assess whether this shift contributed to the decline in U.S. output volatility over the period 1949–2005 and provide an estimate of its relative importance. Growth rates of GDP by industry are analyzed in a seemingly unrelated multivariate autoregression framework with time-varying innovation covariance matrices. These changing unobserved covariance matrices are modeled as a Wishart autoregressive process of order one, which results in a nonlinear state-space system. The particle filter is used to obtain estimates of the innovation covariance matrix at each point in time. Several counterfactual experiments make it possible to apportion the decline in output volatility between the shift in the sectoral composition and changes in innovations. Our main finding is that the shift into the service sector can explain about 30% of the decline in GDP's volatility, despite the fact that some sectors became even more volatile. This result is robust across a wide variety of alternative specifications.

Measuring Autonomy and Emergence via Granger Causality

Concepts of emergence and autonomy are central to artificial life and related cognitive and behavioral sciences. However, quantitative and easy-to-apply measures of these phenomena are mostly lacking. Here, I describe quantitative and practicable measures for both autonomy and emergence, based on the framework of multivariate autoregression and specifically Granger causality. G-autonomy measures the extent to which the knowing the past of a variable helps predict its future, as compared to predictions based on past states of external (environmental) variables. G-emergence measures the extent to which a process is both dependent upon and autonomous from its underlying causal factors. These measures are validated by application to agent-based models of predation (for autonomy) and flocking (for emergence). In the former, evolutionary adaptation enhances autonomy; the latter model illustrates not only emergence but also downward causation. I end with a discussion of relations among autonomy, emergence, and consciousness.

Stochastic Precipitation Generation Based on a Multivariate Autoregression Model

The problem of stochastic precipitation generation has long been of interest. A good generator should produce time series with statistical properties to match those of the real precipitation. Here, a multivariate autoregression model designed to capture the covariance and lag-1 cross-covariance structure of the precipitation measurements is presented. A truncated and power-transformed normal distribution is used to simultaneously model both occurrences and amounts of daily precipitation. The methodology is illustrated using daily rain gauge datasets for three areas in the continental United States.

MULTIVARIATE AUTOREGRESSION OF ORDER ONE WITH INFINITE VARIANCE INNOVATIONS

We consider the limiting behavior of a vector autoregressive model of order one (VAR(1)) with independent and identically distributed (i.i.d.) innovations vector with dependent components in the domain of attraction of a multivariate stable law with possibly different indices of stability. It is shown that in some cases the ordinary least squares (OLS) estimates are inconsistent. This inconsistency basically originates from the fact that each coordinate of the partial sum processes of dependent i.i.d. vectors of innovations in the domain of attraction of stable laws needs a different normalizer to converge to a limiting process. It is also revealed that certain M-estimates, with some regularity conditions, as an appropriate alternative, not only resolve inconsistency of the OLS estimates but also give higher consistency rates in all cases.