Global neural dynamics emerge from multi-scale brain structures, with neurons communicating through synapses to form transiently communicating networks. Network activity arises from intercellular communication that depends on the structure of connectome tracts and local connection, intracellular signalling cascades, and the extracellular molecular milieu that regulate cellular properties. Multi-scale models of brain function have begun to directly link the emergence of global brain dynamics in conscious and unconscious brain states to microscopic changes at the level of cells. In particular, AdEx mean-field models representing statistical properties of local populations of neurons have been connected following human tractography data to represent multi-scale neural phenomena in simulations using The Virtual Brain (TVB). While mean-field models can be run on personal computers for short simulations, or in parallel on high-performance computing (HPC) architectures for longer simulations and parameter scans, the computational burden remains high and vast areas of the parameter space remain unexplored. In this work, we report that our TVB-HPC framework, a modular set of methods used here to implement the TVB-AdEx model for GPU and analyze emergent dynamics, notably accelerates simulations and substantially reduces computational resource requirements. The framework preserves the stability and robustness of the TVB-AdEx model, thus facilitating finer resolution exploration of vast parameter spaces as well as longer simulations previously near impossible to perform. Given that simulation and analysis toolkits are made public as open-source packages, our framework serves as a template onto which other models can be easily scripted and personalized datasets can be used for studies of inter-individual variability of parameters related to functional brain dynamics.

翻译：全局神经动力学源于多尺度脑结构，其中神经元通过突触相互通信，形成瞬时通信网络。网络活动依赖于细胞间通信，这种通信受连接组束结构、局部连接、胞内信号级联反应以及调节细胞特性的胞外分子环境的影响。脑功能的多尺度模型已开始将清醒与无意识脑状态下全局脑动力学的涌现直接与细胞层面的微观变化联系起来。特别是，代表局部神经元群统计特性的AdEx平均场模型已结合人类纤维束成像数据，在利用虚拟脑（TVB）的仿真中表征多尺度神经现象。尽管平均场模型可在个人计算机上运行短时仿真，或在高性能计算（HPC）架构上并行运行以进行更长时仿真和参数扫描，但其计算负担仍然较大，且参数空间中大部分区域尚未探索。本文报告了我们的TVB-HPC框架——一组用于在GPU上实现TVB-AdEx模型并分析涌现动力学的模块化方法——能够显著加速仿真并大幅降低计算资源需求。该框架保持了TVB-AdEx模型的稳定性和鲁棒性，从而可对广阔参数空间进行更精细的分辨率探索，并执行此前近乎不可能的更长时仿真。鉴于仿真与分析工具集已作为开源软件包公开发布，我们的框架可作为模板，便于其他模型快速脚本化，并可利用个性化数据集研究功能性脑动力学相关参数的个体间变异性。