Theoretical and Computational Neuroscience
Author: Federico Miceli | Email: micelifederico08@gmail.com
Federico Miceli 1°, Mauro Granado 2°, Nataniel Martinez 3°, Fernando Montani 4°
1° IFLP-UNLP-CONICET
To characterize the nonlinear dynamics of a neural network connecting two stochastic systems, it is necessary to know the flow of information connecting them. A statistical tool commonly used in these studies is the
mutual information between the systems. If a time shift is added to this measure, transition probabilities can be studied to discern the dynamical properties of the network. In particular, this method marginalizes the directional information exchanged between the systems by not filtering the information coming from inputs common to the set or shared by both. However, using a Markov process, an asymmetric quantity can be defined between ensembles of neurons through the transfer entropy, which focuses on quantifying the flow of information that circulates only in a given direction between the two systems.
In this work, information transfer was quantified in intracranial recordings from patients with drug-naïve refractory epilepsy. These calculations were performed in those electrodes involved in the area responsible for the generation of epileptic seizures, called the epileptogenic zone. Since transfer entropy is able to detect the directed exchange of information between two systems, in our case, the channels involved according to electrophysiological reports showed an increase in the rate of information transfer in the moments before the epileptic seizure, with respect to a basal temporal recording far before the seizure. Thus, this tool of information theory