The unprecedented scientific response to the SARS-CoV-2 pandemic involved researchers from diverse fields, resulting in a massive body of research by January 2021 (over 166,000 papers). This study analyzes this research to understand the scale and nature of collaboration. Collaboration in scientific research is well-documented to increase research impact, although coordination costs can be high. While some studies indicated increased novel collaboration during the pandemic, others suggested less international collaboration than expected and smaller team sizes compared to pre-2020 research. This study aims to evaluate the scale and nature of collaboration in COVID-19 research during 2020, using scientometric analysis to compare COVID and non-COVID publications before and during the pandemic. The study uses three collaboration measures: the Collaboration Index (CI), author multidisciplinarity, and team multidisciplinarity, to assess the effects of the pandemic on collaboration.

Existing literature highlights the benefits of collaboration and the positive correlation between research team size and impact. Multidisciplinary research is recognized for its success in addressing complex challenges. Numerous initiatives promoted collaboration during the COVID-19 response, including global research databases, calls for teamwork, and open access publications. However, early studies showed conflicting results on the extent of collaboration during the pandemic, with some suggesting less international collaboration and smaller team sizes compared to pre-pandemic research, possibly due to the coordination costs associated with multidisciplinary research.

The study uses data from the COVID-19 Open Research Dataset (CORD-19), supplemented by the Microsoft Academic Graph (MAG). The dataset consists of 166,356 COVID-19 related articles published in 2020, 4,017,655 non-COVID-19 articles published between 2016 and 2019, and 1,205,434 non-COVID-19 articles published in 2020. The MAG fields of study (FoS) are used to categorize research papers. To evaluate collaboration, three measures were employed: 1. **Collaboration Index (CI):** The ratio of the number of authors of co-authored articles to the total number of co-authored articles. The study generates CI distributions for each dataset by resampling 50,000 papers 1000 times. 2. **Author Multidisciplinarity:** This measures the extent to which authors publish across multiple disciplines. A bipartite network is constructed, linking researchers to subjects based on their publications. The network is projected to produce a graph of sub-disciplines, where edges are weighted according to the number of authors publishing in both fields. The proportion of total edge weights between communities (disciplines) is calculated to assess multidisciplinarity. 3. **Research Team Disciplinary Diversity:** Publication vectors representing authors' research profiles (based on the proportion of their work published across different disciplines) are used to quantify team diversity using a cosine similarity measure. Team disciplinary diversity (1-S_team) is calculated, comparing diversity by team size for COVID-19 and non-COVID-19 teams. Case studies of multidisciplinarity are explored using a modified network structure visualizing the relationships between fields of study in pre-2020 and COVID-19 research. Flow diagrams are created to illustrate these relationships for Virology, Computer Science, Materials Science, and Development Economics.

The study reveals that COVID-19 research teams were significantly smaller than their non-COVID-19 counterparts, exhibiting a lower Collaboration Index (CI) despite an increasing CI trend for non-COVID research. However, COVID-19 research showed higher author multidisciplinarity and more diverse research teams. Author multidisciplinarity, measured as the proportion of edges between disciplines in an author-FoS network, increased significantly in 2020, with a greater increase when including COVID-19 research. COVID-19 research teams were significantly more diverse than pre-2020 and 2020 non-COVID-19 teams of the same size. Case studies using flow diagrams highlighted the multidisciplinary nature of COVID-19 research across various fields, such as Virology, Computer Science, Materials Science, and Development Economics, showcasing collaborations between previously less-connected disciplines. For example, the study showcased collaborations between Computer Science and Medicine in the application of machine learning for COVID-19 detection from medical images.

The findings suggest a trade-off between team size and multidisciplinarity in COVID-19 research. While smaller teams were prevalent, the increased multidisciplinarity indicates that researchers prioritized efficient collaboration across diverse backgrounds to address the urgency of the pandemic. This contrasts with the general trend of increasing collaboration and team size in research. The increased multidisciplinarity in COVID-19 research may be evidence of its disruptive nature, connecting otherwise separate research communities. The smaller team sizes might not necessarily imply reduced impact, as some larger teams produced highly cited publications. The methodology developed in this study can be applied to future scientometric analyses to assess and visualize multidisciplinarity in research.

This study demonstrates that COVID-19 research, while characterized by smaller teams than usual, fostered significantly greater multidisciplinary collaboration. The findings suggest a strategic adaptation to the pandemic's urgency, prioritizing efficient cross-disciplinary collaboration over larger team sizes. This approach may offer valuable insights for future large-scale research initiatives. The novel visualization methods presented can be beneficial to researchers and funding agencies in understanding and promoting multidisciplinary research.

The study relies on the accuracy and completeness of the CORD-19 and MAG datasets. The classification of research fields might be subject to some degree of uncertainty. The analysis focuses on collaboration patterns in publications, which might not fully capture all forms of collaboration (e.g., informal collaborations).