Faculty of Medicine and Health Technology, <a href="https://ror.org/033003e23">Tampere University</a>, 33520 Tampere, Finland.
Phys Rev E. 2024 Oct;110(4-1):044208. doi: 10.1103/PhysRevE.110.044208.
It is well established that neural mass models (NMMs) can effectively simulate the mesoscopic and macroscopic dynamics of electroencephalography (EEG), including epileptic EEG. However, the use of NMMs to gain insight on the neuronal system by parameter estimation is hampered by their high dimensionality and the lack of knowledge on what NMM parameters can be reliably estimated. In this article, we analyze the parameter sensitivity of the Jansen and Rit NMM (JR NMM) in order to identify the most sensitive JR-NMM parameters for reliable parameter estimation from EEG data. We then propose a Bayesian approach for estimating the JR-NMM states and parameters based on an expectation-maximization algorithm combined with the unscented Kalman smoother (UKS EM). Global sensitivity analysis methods including the Morris method and the Sobol method are used to perform sensitivity analysis. Results from both the Morris and the Sobol method show that the average inhibitory synaptic gain, B, and the reciprocal of the time constant of the average inhibitory postsynaptic potentials, b, have a significant impact on the JR-NMM output along with having the least interaction with other model parameters. The UKS-EM method for estimating the parameters B and b is validated using simulations under varying levels of measurement noise. Finally we apply the UKS-EM algorithm to intracranial EEG data from 16 epileptic patients. Our results, both at individual and group level show that the parameters B and b change significantly between the preseizure and seizure period, and between the seizure and postseizure period, with the transition to seizure characterized by a decrease in the average B, and the high frequency activity in seizure characterized by an increase in b. These results establish a sensitivity analysis guided Bayesian parameter estimation as a powerful tool for reducing the parameter space of high-dimensional NMMs enabling reliable and efficient estimation of the most sensitive NMM parameters, with the potential for online and fast tracking of NMM parameters in applications such as seizure tracking and control.
已有研究证实，神经群模型（NMM）可有效模拟脑电（EEG）的介观和宏观动力学，包括癫痫性 EEG。然而，由于 NMM 维度高，且缺乏可对其参数进行可靠估计的相关知识，故其在通过参数估计深入了解神经元系统方面的应用受到阻碍。本文分析了詹森-里特神经群模型（JR-NMM）的参数敏感性，以确定 JR-NMM 中最敏感的参数，从而能够从 EEG 数据中进行可靠的参数估计。然后，我们提出了一种基于期望最大化算法与无迹卡尔曼平滑（UKS EM）相结合的贝叶斯方法，用于估计 JR-NMM 的状态和参数。全局敏感性分析方法包括 Morris 方法和 Sobol 方法，用于进行敏感性分析。Morris 方法和 Sobol 方法的结果均表明，平均抑制性突触增益 B 和平均抑制性突触后电位时间常数 b 的倒数对 JR-NMM 的输出具有显著影响，并且与其他模型参数的相互作用最小。使用不同测量噪声水平下的模拟对用于估计参数 B 和 b 的 UKS EM 方法进行了验证。最后，我们将 UKS EM 算法应用于 16 名癫痫患者的颅内 EEG 数据。个体和组水平的结果均表明，参数 B 和 b 在发作前和发作期间以及发作和发作后期间均发生显著变化，且发作向发作的转变特征是 B 的平均值降低，发作中的高频活动特征是 b 的增加。这些结果确立了一种基于敏感性分析的贝叶斯参数估计方法，作为一种强大的工具，用于降低高维 NMM 的参数空间，从而能够对最敏感的 NMM 参数进行可靠且高效的估计，并有可能在癫痫跟踪和控制等应用中实现 NMM 参数的在线和快速跟踪。
