Epileptic seizure suppression: A computational approach for identification and control using real data.

Area of Science:

• Computational neuroscience
• Biomedical engineering
• Signal processing

Background:

• Epilepsy affects millions globally, necessitating advanced research into its mechanisms and suppression.
• Current computational models for epilepsy have limitations in accurately describing real-world seizure signatures.
• State-space models are useful but restricted to phenomena they can precisely represent.

Purpose of the Study:

• To propose a system identification (SI) approach using autoregressive (AR) models for describing real epileptiform activity.
• To develop a strategy for artificial reconstruction and mitigation of epileptiform activity using non-hybrid and hybrid controllers.
• To investigate the effectiveness of AR models and controllers in attenuating seizures and driving signals to an interictal state.

Main Methods:

• Utilizing autoregressive (AR) modeling, a data-driven system identification (SI) technique, to model epileptiform activity.
• Converting AR models into state-space representations for seizure reconstruction and attenuation.
• Designing and applying non-hybrid and hybrid controllers, derived from ictal and interictal events, respectively.

Main Results:

• Low-order AR models effectively represent complex epileptiform activities.
• Both non-hybrid and hybrid controllers demonstrated significant efficacy in attenuating seizure activity.
• The applied controllers successfully guided the epileptic signals towards a normal, interictal condition.

Conclusions:

• Autoregressive modeling provides a robust method for creating realistic computational models of epilepsy.
• The developed controller strategies offer a promising approach for artificial seizure mitigation.
• Findings support the development of patient-specific models for tailored external stimuli to control seizure activity.