Applying clustering coefficient to the pattern of international author collaboration in neuroimmunology and neuroinflammation

Aim: To apply cluster coefficient (CC) to the pattern of international author collaborations in neuroimmunology and neuroinflammation using data from Medline and to visualize results using Google maps and social network analysis (SNA). Methods: Selecting 2799 abstracts, author names, countries, and keywords on January 22, 2018 from Medline based on keyword neuroimmunology (or neuroinflammation) within the article title since 1982, we reported following features: (1) nation distribution for the 1st author’s nationality; (2) eminent journals and authors in the field of neuroimmunology and neuroinflammation; (3) notable keywords defined by authors representing both neuroimmunology and neuroinflammation; and (4) CCs in networks. We programmed Microsoft Excel VBA routines to extract data from Medline and used Google Maps and SNA Pajek software to display graphical representations with an easy-to-read feature for readers. Results: We found that: (1) the most number of papers in neuroimmunology and neuroinflammation are from the USA (902, 32.23%) and China (363, 12.97%); (2) the productive journals and authors in neuroimmunology and neuroinflammation are J Neuroinflammation and PLoS One, and Michael T. Heneka (Germany) and Richard M. Ransohoff (USA); (3) the most linked keywords are interleukin (IL), IL-1beta, and blood brain barrier; (4) author networks present higher CC than those nation networks. Conclusion: SNA provides wide and deep insight into the relationships among nations in co-author collaboration. The results can help readers in future submission to a journal in neuroimmunology and neuroinflammation.


INTRODUCTION
Neuroimmunology is a field combining neuroscience, the study of the nervous system, and immunology in a review of the immune system.Neuroimmunologists seek to better understand the interactions of these two complex systems during the development of homeostasis and response to injuries.A long-term goal is to further develop our understanding of the pathology of certain neurological diseases [1] .
Similarly, neuroinflammation is inflammation of the nervous tissue.It is initiated in response to a variety of cause such as infection, traumatic brain injury, toxic metabolites, or autoimmunity [2] .The central nervous system is typically an immunologically privileged site because peripheral immune cells are blocked by the blood-brain barrier [3,4] .However, the issues including author collaboration and keyword defined by authors are still unclear.
By January 22, 2018, more than 18,282 papers have appeared on PubMed when searching with the keyword (neuroimmunology OR neuroinflammation) and 2885 in the paper title only including the keyword (neuroimmunology [Title] OR neuroinflammation [Title]) since 1982.The issue of which nations dominate the papers published in neuroimmunology and neuroinflammation intrigues us to investigate which keywords are most frequently seen in those articles in the past decades.
Big data is a concept that has evolved from the modern trend of scientism [5] .Many data scientists develop ways to discover new knowledge from the vast quantities of increasingly available information [5] .Even an apocryphal story was often told to explain the concept of co-occurrence between beer and diaper sales [6][7][8] , the way of finding both beer and diaper sales had a strong correlation on Friday is unclear in the literature.All possible pairs of observed goods or services are worth studying the association, which is similar to the keywords and authors in journal papers.Fortunately, social network analysis (SNA) [9][10][11] can analyze big data for us to investigate the association of any pairs of goods or services in a network.
Authorship collaboration using SNA is an example regarding co-authors in recent years [9] because coauthors among researchers form is a type of social network.Whether the authoring network earns a higher centrality measure (or density) than the national system is required to explore.We are thus interested in using SNA to explore the features in neuroimmunology and neuroinflammation from published papers we observed in Medline library.
Google Maps provide an overall view of geospatial visualization with coordinates of latitude and longitude on a map [12,13] .However, a few appeared in Medline library in search of keyword "google map [Title]" on November 22, 2017.Many papers [9][10][11] have studied on co-author collaboration in academics, however, none display results with Google Maps and SNA.
Our aims are to apply the clustering coefficient [14] to the pattern of international author collaboration in neuroimmunology and neuroinflammation on the following topics: (1) nation distribution; (2) the most eminent journals and authors in neuroimmunology and neuroinflammation; (3) the recent research domains defined by authors; and (4) the cluster coefficients (CCs) in different networks.

Data sources
We programed Microsoft Excel VBA modules to extract abstracts and their corresponding co-author names as well as author-defined keywords for each article on January 22, 2018, from Medline since 1982.Only those abstracts published by the keyword (neuroimmunology [Title] OR neuroinflammation [Title]) were included.Others like those labeled with Published Erratum, Editorial or without author nation name were excluded from this study.A total of 2885 eligible abstracts were obtained from the Medline.

Data arrangement to fit SNA requirement
Before visualizing our results using SNA, we organized data in compliance with the format and guidelines defined by Pajek software [15] .Microsoft Excel VBA was used to deal with data fitting to the SNA requirement.

Graphical representations to report
Author nations and their relations Two tables (i.e.columns for publication years and rows for the 1st author nations or journals) were made for presenting the distribution in nation (of journal) for the domain of neuroimmunology and neuroinflammation.The bigger bubble means, the more number of the nodes (i.e.nations, or keywords in this study).The wider line indicates, the stronger relations between the 2 nodes.Community clusters are filled with different colors in bubbles.The most eminent authors were calculated from the Medline library since 1982.

Statistical tools and data analyses
Google Maps [16] and SNA Pajek software [15] were used to display and visualized representations for eminent authors and keywords in relation with neuroimmunology and neuroinflammation.Author-made Excel VBA modules were applied to organize the data.CC represents the density of a network and a significant level (> 1.96) is defined by t-value as the formula [= CC × [(n-1)/(1-CC 2 )] 1/2 ], where n = the number of nodes in a network.
In contrast, E-I index is defined by the formula, where EL = the number of external friendship links and IL = the number of internal friendship links [17] .The negative E-I index means a coherence cluster in existence.Similarly, the higher CC indicates many members are members' friends linked to others.Density denotes as the ratio of the linkage members over all possible members.

Author nations and their relations
A total of 2799 eligible papers with complete author nations based on journal article since 1982 are in Table 1.We can see that the most number of articles are from the countries of USA (902, 32.23%) and China (363, 12.97%).The trend in the number of publications with authorship is present in the column of growth in Table 1.All nations but the Netherlands and Israel showed a positively increase.
The diagram in Figure 1 displays author collaboration among nations.Overall, the highest production in countries are from the USA, China, and Europe [Figure 1].Any collaborated with others are shown with a blue line.Interested authors are recommended to click the bubble of interest to see details on a website at the reference [18] .

The eminent authors in surgery
The most prominent journals and authors with the most number of papers in neuroimmunology and neuroinflammation are J Neuroinflammation and PLoS One as well as Michael T. Heneka (Germany) and  Richard M. Ransohoff (USA) [Table 2 and Figure 2].The link on website appears at http://www.healthup.org.tw/gps/Neuroimmunologyauthor.htm [19].

Keywords to present the research domain
The most linked keywords are interleukin (IL), IL-1beta, and blood-brain barrier [Figure 3] [20] .We can see that the keywords consisting of many clusters with different cluster coefficients.

Cluster coefficients in a network
Each cluster has its cluster coefficient representing the density of a network.We found that author clusters earn higher CC than have nation clusters.Cluster coefficient has a significant effect in comparison with a considerable t-value (> 2.0), indicating author network with more significance than those of the nations [Table 3].

Figure 1 .
Figure 1.Google Maps on the topic of author collaboration in neuroimmunology and neuroinflammation (cluster coefficient = 0.61)

Table 3 . CCs in each cluster and their relevant indicators Name CC Density Weighted EI Node Degree Weighted t -value
Denotes significance when t -value greater than 2.0.CC: cluster coefficient. *