7th International Conference on Engineering and Natural Sciences (ICENS 2021), Sarajevo, Bosnia And Herzegovina, 23 - 27 July 2021, pp.60-66
The model for
controlling epilepsy discussed here is based on the seizures suppression
experimental methods via the electrical stimulation of brain. It has a
potential of “fine tuning” according to the epileptic pathology specifics of
patients. We consider here a simplified case of an artificial neural network
(ANN) with the Hodgkin-Huxley elements providing the necessary variety of
dynamical regimes: individual neuron spikes and bursts which could cause the
hyper-synchronized behavior of epileptiform type in the whole network. We
perform a fine control of the ANN dynamics with a single element which plays
two roles: it detects the coming seize and send a feedback signal to other
neurons to suppress the epileptiform dynamics. To increase the quality and efficiency
of the control we study non-classical
(based on the quantum paradigm) algorithm. Recently we demonstrated the
ability of a single Hodgkin-Huxley neuron to emulate some quantum classification and searching algorithms
in a relatively profitable way. Here we reproduce our approach to detect and
suppress epileptiform dynamics in the small ANN. The feedback loop to other
neurons could be based on optimal / suboptimal gradient approaches or on an
artificial attractor forming in the dynamical system. We study the efficiency
and robustness of our proposed algorithm and discuss its pros and cons to
compare with our recent classical algorithm-based model of the epileptiform
suppression.