Analysis of Microbiome Research using Co-Author Network

Abstract

To identify the structure and causes of scientific research growth, we performed bibliographic analysis and network analysis. We selected the field of microbiome research, which has recently experienced rapid development and has a substantial research scale. Specifically, we choose the top 11 countries in terms of global microbiome research scale and analyzed the network of these countries from 2000 to 2021 to identify commonalities and differences in research network changes per country.

Through bibliographic analysis, we confirmed that the rapid growth in the microbiome field began in the early 2010s. The growth timing of most countries was behind that of the United States. Of interest, we observed a phenomenon of research scale reversal between the U.S. and China. By conducting bibliographic analysis, we were able to identify consistent growth patterns, enabling us to make predictions about the expansion of science networks. To further investigate this approach, we constructed a Co-Author Network, which represents interconnections among scientists and plays a pivotal role in the scientific research network. By analyzing the quantity of network nodes that represent authors, we achieved the ability to forecast network growth. The growth of networks in the majority of countries closely aligned with the predictions derived from bibliographic analysis.

Using various metrics from network theory, we examined the structure of the network. Initially, we looked at the Average Clustering Coefficient (ACC), which quantifies the cohesiveness among adjacent nodes. The scientific research network exhibited a high ACC value from its inception, gradually decreasing over time. This suggests that researchers within the network maintained strong connections throughout its growth, albeit slightly decreasing over time. Notably, we observed a faster decline in ACC within the Chinese network compared to other countries.

The Average Path Length (APL), which indicates network information efficiency, displayed a unique pattern. While an increase in APL is typically associated with a growing number of network nodes, most networks showed an S-shaped increase that eventually converged to a certain value. However, in certain countries, APL followed a linear increase and also converged. This finding highlights the previously unknown phenomenon of APL convergence during network creation and development. Additionally, our research demonstrated the applicability of network creation models such as the "Erdős-Rényi model" and the "Watts-Strogatz model" to real-world network cases. Based on our results, we predict that the microbiome research field will approximate a "Small World network," characterized by a high ACC and short APL, akin to the Watts-Strogatz model. Particularly during its early stages, the network closely approximates a Small World network and continues to do so as it grows. However, if the ACC gradually decreases, as observed in APL convergence at a certain distance, the continuous approximation to the Small World network may be compromised. The rapid decline of ACC, as seen in China, could be interpreted as an example of the swift randomization of research networks. According to the "Erdős-Rényi model," a giant component, representing the largest connected component in the entire network, tends to emerge in sufficiently large and dense networks. Throughout the growth period of the 11 countries, all of them exhibited giant components. However, we also observed that the proportion of giant components did not directly correlate with changes in network size. This observation implies the differences in network structure development among countries.

Next, we examined the characteristics of key nodes in each network to elucidate the factors driving network growth and scientific progress. Specifically, we analyzed Betweenness centrality, which quantifies the number of times a node serves as a bridge between different researchers. The status of key nodes in each network showed dynamic changes as the network developed. Notably, the top nodes in the network did not maintain a consistent position in terms of connectivity over the analysis period, indicating a dynamic nature of scientific network development. In the cases of the United States and China, nodes related to technology emerged as rising stars within their respective networks and demonstrated connections with other country networks. This highlights the significant influence of technology among the factors driving network development.