用基于模拟的推断量化脉冲神经元网络中的过度兴奋代偿机制
Compensation of hyperexcitability with simulation-based inference
一项发表于 eLife 的研究用基于模拟的推断（simulation-based inference）在脉冲神经元网络模型中量化了多种假定代偿机制的相互作用。模型中的多个参数可以相互补偿以维持基线活动，研究估计了各自的代偿潜力。针对过度兴奋的不同成因——中间神经元丢失、兴奋性递归突触和主细胞去极化——存在可恢复正常兴奋性的不同代偿机制。结果表明，脉冲神经元网络模拟器能够为病理生理网络机制生成假设。
来源：eLife · doi.org
The activity of healthy neuronal networks is tightly regulated, and a shift toward hyperexcitability can cause various problems, such as epilepsies, memory deficits, and motor disorders. Numerous cellular, synaptic, and intrinsic mechanisms of hyperexcitability and compensatory mechanisms to restore healthy activity have been proposed. However, quantifying multiple compensatory mechanisms and their dependence on specific pathophysiological mechanisms has proven challenging, even in computational models. We use simulation-based inference to quantify the interactions of putative compensatory mechanisms in a spiking neuronal network model. Various parameters of the model can compensate for the changes in other parameters to maintain baseline activity, and we estimated their compensatory potential. Furthermore, specific causes of hyperexcitability – interneuron loss, excitatory recurrent synapses, and principal cell depolarization – have distinct compensatory mechanisms that can restore normal excitability. Our results show that spiking neuronal network simulators could generate hypotheses about the mechanisms of pathophysiological network mechanisms.