ABSTRACT

Neuropsychological tests have commonly been used to determine the organization of cognitive functions by identifying latent variables. In contrast, an approach which has seldom been employed is network analysis. We characterize the network structure of a set of representative neuropsychological test scores in cognitively healthy older adults and MCI and dementia patients using network analysis. We employed the neuropsychological battery from the National Alzheimer’s Coordinating Center which included healthy controls (n = 7623), mild cognitive impairment patients (n = 5981) and dementia patients (n = 2040), defined according to the Clinical Dementia Rating. The results showed that, according to several network analysis measures, the most central cognitive function is executive function followed by attention, language, and memory. At the test level, the most central test was the Trail Making Test B, which measures cognitive flexibility. Importantly, these results and most other network measures, such as the community organization and graph representation, were similar across the three diagnostic groups. Therefore, network analysis can help to establish a ranking of cognitive functions and tests based on network centrality and suggests that this organization is preserved in dementia. Central nodes might be particularly relevant both from a theoretical and clinical point of view, as they are more associated with other nodes, and their disruption is likely to have a larger effect on the overall network than peripheral nodes. The present analysis may provide a proof of principle for the application of network analysis to cognitive data.

 抽象的

神经心理学测试通常用于通过识别潜在变量来确定认知功能的组织。相比之下，很少采用的方法是网络分析。我们使用网络分析来描述认知健康的老年人以及轻度认知障碍和痴呆症患者的一组代表性神经心理学测试分数的网络结构。我们采用了国家阿尔茨海默病协调中心的神经心理学电池，其中包括健康对照（n = 7623）、轻度认知障碍患者（n = 5981）和痴呆患者（n = 2040），根据临床痴呆评级进行定义。结果表明，根据多项网络分析指标，最核心的认知功能是执行功能，其次是注意力、语言和记忆。在测试层面，最核心的测试是 Trail Making Test B，它衡量认知灵活性。重要的是，这些结果和大多数其他网络测量（例如社区组织和图形表示）在三个诊断组中是相似的。因此，网络分析可以帮助建立基于网络中心性的认知功能和测试的排名，并表明该组织在痴呆症中得以保留。从理论和临床的角度来看，中心节点可能特别相关，因为它们与其他节点的关联性更强，并且它们的破坏可能比外围节点对整个网络产生更大的影响。本分析可以为网络分析应用于认知数据提供原理证明。