A feedback control systems view of epileptic seizures
Cybernetics and Systems Analysis
Regularization in the selection of radial basis function centers
Neural Computation
| Hi-index | 0.01 |
This paper explores a computational dynamic solution for epileptic seizure control. A neuronal model of epileptiform activity which is driving to periodicity, simulating seizures, is chaotified via two chaos anti-control algorithms. Anti-control of some in vivo measurements is also put to test. The perspective is that in an epileptic disorder, a transition of the state of the neuronal system from being chaotic to being patho-physiologically periodic can cause this dynamical disorder. The aim is to retrieve the chaotic state which was running before the seizure start. Based on identifying loss regions for period orbits in the state space, the first control method acts by avoiding the system trajectories entrance into the defined loss regions. On the other hand, the second method, based on intra-attractor (orbit) control concept, is an attempt to steer trajectories to the unstable manifold of a target unstable period orbit solution embedded within the system dynamic. The presented anti-control methods make use of a novel strategy to model chaotic time series' return map with radial basis function to improve the estimation of the dynamics under investigation. Our results show that both methods are successful in driving the model's dynamical activity away from the periodic direction while the second one seems to be faster and more efficient. Moreover, testing both methods on some hippocampal recordings of epileptized rats yields fairly acceptable results in dynamic modulation. This opens ways for further online experimental stimulations based on chaos concept.