Performance of closed-loop control of epileptiform spikes in neural mass models

Abstract

Neural mass models can produce electroencephalography (EEG) like signals corresponding to interical, pre-ictal and ictal activities. In this paper, a novel closed-loop feedback control strategy based on algebraic estimation is proposed to eliminate epileptiform spikes in neural mass models. Algebraic estimation plays a role in observing the states of the model in order to construct the controller. For a network of coupled neural populations, the characteristics regarding the closed-loop feedback control strategy, including the relationship between the type of controlled populations and the ability of eliminating epileptiform spikes, the relationship between the number of controlled populations and control energy, the relationship between the model parameters and control energy, are determined by numerical simulations. The purpose is to establish the rules for the proper control of eliminating epileptiform spikes with as less control energy as possible. Moreover, the proposed control-loop control strategy is compared with a direct proportional feedback control strategy by numerical simulations. It is shown that the use of algebraic estimation makes a reduction of control energy.