Event Abstract

Reassessing Aphasic Classifications with Community Detection Analysis

  • 1 Drexel University, Psychology, United States
  • 2 University of Alabama at Birmingham, Psychology, United States

Traditional models of aphasia have emphasized the distinction between speech production and comprehension deficits. This distinction dates back to the 1800’s and remains foundational to the aphasia sub-typing framework used in clinical aphasiology today. However, many patients do not fit into the classic aphasia sub-types (Caplan, 2012; Kasselimis et al., 2017). The aggregation of data from people with aphasia in large, publicly-available databases, combined with the development of novel analytic techniques, provides a new opportunity for data-driven discovery of aphasia sub-types. In the current study, community detection analysis was used to identify clusters of individuals with aphasia who had similar behavioral deficit profiles. Community detection analysis attempts to uncover groups of densely connected nodes in a network that are sparsely connected to other groups. The data came from the Moss Aphasia Psycholinguistics Project Database (Mirman et al., 2010) and consisted of 21 test scores from 296 participants with aphasia. The scores included general aphasia assessment (WAB), tests of short-term memory, sentence comprehension, semantic processing, lexical processing, and speech perception and production. Participants missing more than 70% of their data were excluded from the analysis (n=65) and multiple imputations were used to fill in the remaining missing data, resulting in five complete data sets. To perform the analysis, correlations across the 21 test scores were computed for all pairs of participants. Network graphs were constructed from thresholded correlation matrices such that there were no isolated participant nodes. Community detection analyses were performed using the edge-betweenness algorithm (Girvan & Newman, 2002) on each of the five graphs. Majority community assignment was used to combine the results from the five graphs: if a participant was placed in a community in at least three out of the five graphs, then they were assigned to that community. Almost all of the participants (96.9%) were placed into one of the three main communities. Nine participants (3.9%) were not consistently placed into one of these communities – substantially lower than the 26.5% (13 out of 49) “unclassified” aphasia rate reported by Kasselimis et al (2017). These three communities did not align with the traditional classifications of aphasia. Community one (N=111) consisted of individuals with relatively mild aphasia (WAB AQ Mean = 85.5, range = 58.8-99.3), who had generally better scores on all tests. Many of them (65.8%) fell into the “Anomic” aphasia sub-type, though this community also included a substantial number of people with the Broca’s (15.3%) and Conduction (15.3%) sub-types. Communities two and three had similar severity scores, but participants in community two (N=47) had phonological deficits (e.g., poor rhyme discrimination and high rates of formal errors in picture naming) and participants in community three (N=64) had semantic deficits (e.g., low Camel and Cactus Test scores and high rates of semantic errors in picture naming). Although the distinction between speech production and comprehension is important, these results suggest that it is not the primary one. Rather, after overall impairment severity, the primary distinction appears to be between phonological and semantic deficits, affecting both speech production and comprehension.

Figure 1

References

Caplan, D. (2012). Aphasic Syndromes. Clinical Neuropsychology.

Girvan, M., & Newman, M. E. J. (2002). Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the United States of America, 99(12), 7821–6. http://doi.org/10.1073/pnas.122653799

Kasselimis, D. S., Simos, P. G., Peppas, C., Evdokimidis, I., & Potagas, C. (2017). The unbridged gap between clinical diagnosis and contemporary research on aphasia: A short discussion on the validity and clinical utility of taxonomic categories. Brain and Language, 164, 63–67. http://doi.org/10.1016/j.bandl.2016.10.005

Mirman, D., Strauss, T. J., Brecher, A., Walker, G. M., Sobel, P., Dell, G. S., & Schwartz, M. F. (2010). A large, searchable, web-based database of aphasic performance on picture naming and other tests of cognitive function. Procedia - Social and Behavioral Sciences, 6(6), 132–133. http://doi.org/10.1016/j.sbspro.2010.08.066

Keywords: Aphasia diagnosis, Classification, Sub-typing, Community Detection Analysis, Network analysis

Conference: Academy of Aphasia 55th Annual Meeting , Baltimore, United States, 5 Nov - 7 Nov, 2017.

Presentation Type: poster or oral

Topic: Consider for student award

Citation: Landrigan J and Mirman D (2017). Reassessing Aphasic Classifications with Community Detection Analysis. Front. Hum. Neurosci. Conference Abstract: Academy of Aphasia 55th Annual Meeting . doi: 10.3389/conf.fnhum.2017.223.00032

Received: 27 Apr 2017; Published Online: 11 Dec 2017.

* Correspondence: Mr. Jon-Frederick Landrigan, Drexel University, Psychology, Philadelphia, PA, 19104, United States, jon.landrigan@gmail.com

© 2007 - 2018 Frontiers Media S.A. All Rights Reserved