A Review of The Dynamical Systems Modeling of Epileptic Seizures For Onset Prediction

Abstract

In this review, we present a critical appraisal of works that consider epilepsy as a dynamic disease and therefore presentable from the perspective of dynamic system theory. Epilepsy is an acute brain disorder characterized by recurrent seizures where parts of the brain elicit abnormally synchronous electrical activity. The most commonly encountered forms of epilepsy are generalized absence epilepsy and temporal lobe epilepsy. The electroencephalography (EEG) which is the recording of the fluctuating electric field of the brain is the major clinical diagnostic tool for epilepsy and also a vital source of data for epilepsy research. In majority of cases accurate diagnosis of the disease can be made and seizures are controlled by the regular use of anti-epileptic drugs (AEDs). However, approximately 30% of epileptic patients suffer from medically refractory epilepsy which has defied all existing treatment protocols. Understanding the mechanisms underlying these forms of epileptic seizures and the development of alternative effective treatment strategies is a fundamental challenge in modern epilepsy research. Experimental researches show that the mechanisms involved in refractory epilepsy are so diverse and complex that it is a formidable task to obtain a single framework that categorizes all the pathophysiological changes in the properties of the epileptic brain involved. There has evolved massive evidence that seizures do not occur abruptly as it has been earlier thought but develop over time even hours before the clinical symptoms, thus raising the hope for predictability of epileptic seizure occurrence. Thus, models of the epileptic brain can be postulated using concepts from deterministic and nondeterministic dynamical systems modelling. The main idea is that since the epileptic brain transitions into and out of seizures we can view it as a dynamical system. The deterministic and non-deterministic models are based on seizure onset detection algorithm for the design of a closed loop seizure warning/intervention system. The major focus being the stimulation of the epileptic brain by sending electrical pulses to it in order to disrupt seizure progression once its onset has been detected. Finally, we considered the essential issues in epileptic seizure prediction including the sceptism expressed in recent publications on the validity of nonlinear dynamical systems modelling to epileptic seizure prediction.

Keywords

• epilepsy
• electroencephalogram
• dynamical systems theory
• computational modeling
• deterministic models
• non deterministic models
• neural mass models
• adaptive neuro-fuzzy inference system
• seizure prediction