Research On The Modeling And Control Of Epileptic State In Hippocampus

Epilepsy characterized by abnormal synchronous discharges in large ensembles of neurons is one of the most common neurological diseases.The diagnosis and treatment of epilepsy is a hotpot of neuroscience that many researchers focus on.The traditional open-loop electrical stimulation is an effective treatment for epilepsy,but has poor flexibility and adaptability.Studies show that hippocampus is an important seizure focus and an optional electrical stimulation target for epilepsy.In this paper,the models of hippocampal neuronal network and deep EEG activity were constructed.Then the model predictive control strategy was proposed to suppress the epileptic discharge activity.Firstly,the network of hippocampal CA3 area was established using Pinsky-Rinzel(PR)model to simulate the epileptic and normal network activity.The NARX neural network model of hippocampal activity was obtained based on the simulated input-output data and then used for the prediction of epileptiform system output,thus making preparation for the subsequent design of controller.Secondly,the neural mass model was regarded as a virtual patient with epilepsy to simulate different epileptiform signals used in the subsequent system identification procedures.With the obtained input-output data,the autoregressive PDM models were eventually constructed for different types of epileptic activity.The associated nonlinear functions(ANF)corresponding to the respective PDMs were analyzed.And results demonstrated that some ANFs exhibited evident differences among different states of epilepsy,thus are likely to be features used for the differentiation of epileptic states.Finally,the predictive control approach was combined with the established NARX neural network model to suppress the epileptic seizures.System output was predicted by the NARX model and the model predictive control method was utilized to design stimulation patterns applied on the hippocampus.Results displayed that the epileptic seizure activity was effectively controlled by the closed-loop stimulation strategy.The research provides instruction for the automatic diagnosis of epilepsy and the design of closed-loop electrical stimulation system.Our work also make contributions to the hardware implement and clinical experiment in the future.