Epileptic seizures represent abnormal periods of increased excitation and synchronized bursting of large networks of neurons. While various in vitro models for epilepsy have provided important insights into cellular, synaptic, and local network behavior, additional models are needed to understand recurrent bursting in more characteristic large neural networks. Unfortunately intact animal models have significant limitations for these studies. In contrast, model neural networks offer a number of distinct advantages for studies of large network behavior, including the ability to exactly model and monitor specific neuronal assemblies with known structure, connectivity and membrane properties. Such model neural networks are now being developed and utilized to study recurrent bursting and seizure propagation. Using reduced compartment neurons based on modified Av-Ron Rinzel equations, networks can be assembled that can simulate recurrent bursting behavior and seizure propagation. Modeling is greatly facilitated by distributed computing with computer clusters, allowing for studies on network arrays of 10 6 or more neurons. Since recurrent seizures produce changes in intrinsic neuronal connectivity, the influences of these changes will be studied through specific models with varying connectivity. The ability to disrupt or terminate this bursting behavior with external excitatory stimuli will be modeled and characterized. The importance of feedback loops in burst termination and the specific parameters of external stimulation that optimize burst termination or disruption will be defined. Both discrete external excitatory pulses and low level external excitatory stimulation will be studied. Brain stimulation for the amelioration or control of epileptic seizures in humans is attracting considerable and growing interest. Unfortunately the optimal parameters for favorable modulation by external excitatory stimulation are not well understood and the applications at present use somewhat arbitrary parameters. The studies proposed here may provide important insights into the important criteria responsible for burst generating in large neuronal arrays. From these results the parameters for application of excitatory stimulation for reduction of repetitive bursting in these models networks can be addressed. Hopefully these insights can be applied to devices useful to patients with uncontrolled epilepsy.
