Singular Parameter Prediction Algorithm for Bistable Neural Systems.

An algorithm is presented to predict the intensity and timing of a singular single stimulus required to switch the state of a bistable system from repetitive activity to a stable point. The algorithm is first tested on a modified Hodgkin-Huxley model to predict the parameters of a stimulus capable of annihilating the spontaneously occurring repetitive action potentials. Elevation of the potassium equilibrium potential causes oscillations in the V, m, h and n parameters and generates periodic activity. Equations describing the time-varying behavior of these parameters can be used to predict the pulse width, coupling interval and intensity of a single anodic pulse applied between two consecutive action potentials to suppress the activity. The algorithm was then applied to predict the singular parameters of quasi-periodic epileptiform activity generated in the hippocampus slice preparation exposed to high-potassium concentrations. The results indicate that a stimulus with the estimated parameters was able to either completely annihilate the action potentials in the HH model or predict the region of unpredictable latencies. Therefore this algorithm is capable a predicting singular parameters accurately when the model is known. In the case of an experimental system where the equations of the system are not known, the algorithm predicted parameters in the range of those observed experimentally. Therefore, the algorithm could reduce significantly the amount of time required to find the singular parameters of experimental bistable systems normally obtained by a systematic exploration of the parameter space. In particular, this algorithm could be useful to predict the singular parameters of quasi periodic epileptiform activity leading to the suppression of this activity if the system is bistable.