A Hierarchical Latent Space Network Model for Population Studies of Functional Connectivity
A Hierarchical Latent Space Network Model for Population Studies of Functional Connectivity
复制标题
DOI：
10.1007/s42113-020-00080-0
复制
发表时间：
2020-03
期刊：
影响因子：
--
通讯作者：
James D. Wilson;Skyler J. Cranmer;Zhonglin Lu
中科院分区：
文献类型：
--
作者：
James D. Wilson;Skyler J. Cranmer;Zhonglin Lu
网络神经科学的一个主要挑战在于理解大脑在不同空间尺度上的组织原理。大脑是高度模块化的，因为大脑区域自然地划分为密集连接的子网络，这些子网络本身往往包含密集连接的子网络。对这些复杂的层次结构进行建模是目前阻碍该领域进展的主要技术挑战。我们开发了分层潜在空间模型(HLSM)，该模型可以同时捕捉个体和群体水平的层次结构，解释了功能连通性的多个预测因子，并解释了群体中表现出的个体异质性。我们将我们模型的几个规格应用于从生物医学研究中心卓越项目收集的健康和偏执型精神分裂症患者。我们发现，对于健康人群和患者群体，半球和功能子网络中两个区域的空间位置强烈影响他们连接的倾向。我们还发现，单独而言，两个区域之间的空间距离与它们的连接概率显著负相关，但一旦控制了半球和子网络的位置，它就不再显著了。HLSM还发现健康个体的连通性在患者组中增加了异质性，这表明两个人群之间的总体连通性模式不同。
A major challenge in network neuroscience lies in understanding the organizational principles of the brain at different spatial scales. The brain is highly modular, in that brain regions naturally divide into densely connected subnetworks, which often themselves contain densely connected subnetworks. Modeling these complex hierarchies is a major technical challenge currently inhibiting progress in the field. We develop the hierarchical latent space model (HLSM) that can capture hierarchy at both the individual and population levels, account for multiple predictors of functional connectivity, and account for individual heterogeneity that manifests over a population. We apply several specifications of our model to healthy and paranoid schizophrenia patients collected from the Center for Biomedical Research Excellence project. We find that for both healthy and patient groups, the spatial location of two regions, in hemisphere and functional subnetwork, strongly influence their propensity to connect. We also find that alone, the spatial distance between two regions is significantly and inversely related to their connection probability, but that it is no longer significant once hemisphere and subnetwork locations have been controlled for. The HLSM also identifies increased heterogeneity in the connectivity of the healthy individuals over the patient group, suggesting a difference in overall connectivity patterns between the two populations.