Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dynamical evolution of social network polarization and its impact on the propagation of a virus

Ixandra Achitouv, David Chavalarias

TL;DR

This study analyzes how polarization around vaccination, inferred from Twitter networks during 2020–2022, shapes the diffusion of a virus. Using three real, bidirectional retweet networks labeled as pro- or anti-vaccine via clustering, the authors quantify evolving network properties (density, degree distribution, clustering, assortativity) and observe increasing polarization with cohesive anti-vaccine communities. They then couple these networks to an agent-based transmission model with vaccine effects, comparing polarized (vaccines allocated to pro-vaxx only) versus homogeneous vaccine distribution. Results show polarization accelerates and amplifies viral spread among unvaccinated individuals, with attack rates roughly 2.4× higher in early years, while vaccination within pro-vaxx clusters can screen transmission, underscoring the need to integrate opinion dynamics into epidemic forecasting and public health planning. The work provides a real-network, data-driven framework to study the coupled dynamics of social polarization and disease spread, with practical implications for vaccination strategies and misinformation mitigation.

Abstract

The COVID-19 pandemic that emerged in 2020 has highlighted the complex interplay between vaccine hesitancy and societal polarization. In this study, we analyse the dynamical polarization within a social network as well as the network properties before and after a vaccine was made available. Our results show that as the network evolves from a less structured state to one with more clustered communities. Then using an agent-based modeling approach, we simulate the propagation of a virus in a polarized society by assigning vaccines to pro-vaccine individuals and none to the anti-vaccine individuals. We compare this propagation to the case where the same number of vaccines is distributed homogeneously across the population. In polarized networks, we observe a significantly more widespread diffusion of the virus, highlighting the importance of considering polarization for epidemic forecasting.

Dynamical evolution of social network polarization and its impact on the propagation of a virus

TL;DR

This study analyzes how polarization around vaccination, inferred from Twitter networks during 2020–2022, shapes the diffusion of a virus. Using three real, bidirectional retweet networks labeled as pro- or anti-vaccine via clustering, the authors quantify evolving network properties (density, degree distribution, clustering, assortativity) and observe increasing polarization with cohesive anti-vaccine communities. They then couple these networks to an agent-based transmission model with vaccine effects, comparing polarized (vaccines allocated to pro-vaxx only) versus homogeneous vaccine distribution. Results show polarization accelerates and amplifies viral spread among unvaccinated individuals, with attack rates roughly 2.4× higher in early years, while vaccination within pro-vaxx clusters can screen transmission, underscoring the need to integrate opinion dynamics into epidemic forecasting and public health planning. The work provides a real-network, data-driven framework to study the coupled dynamics of social polarization and disease spread, with practical implications for vaccination strategies and misinformation mitigation.

Abstract

The COVID-19 pandemic that emerged in 2020 has highlighted the complex interplay between vaccine hesitancy and societal polarization. In this study, we analyse the dynamical polarization within a social network as well as the network properties before and after a vaccine was made available. Our results show that as the network evolves from a less structured state to one with more clustered communities. Then using an agent-based modeling approach, we simulate the propagation of a virus in a polarized society by assigning vaccines to pro-vaccine individuals and none to the anti-vaccine individuals. We compare this propagation to the case where the same number of vaccines is distributed homogeneously across the population. In polarized networks, we observe a significantly more widespread diffusion of the virus, highlighting the importance of considering polarization for epidemic forecasting.
Paper Structure (20 sections, 7 equations, 5 figures, 6 tables)

This paper contains 20 sections, 7 equations, 5 figures, 6 tables.

Table of Contents

  1. Introduction
  2. The network dynamics of pro-vaccines vs anti-vaccines
  3. Data collection and analysis of the communities
  4. Network properties analysis
  5. Density
  6. Degree Distribution
  7. Clustering Coefficient (CC)
  8. Assortative mixing in networks
  9. Sociological interpretation
  10. Impact of polarization in the diffusion of a virus
  11. The agent based model and transmission probability
  12. The vaccine scheme
  13. Results
  14. Discussion
  15. Appendices
  16. ...and 5 more sections

Figures (5)

  • Figure 1: Visualization of social network on bidirectional retweets among users sharing COVID related tweets. Cold colors correspond to pro-vaccines users and warm colors to anti-vaccines users. Visualization have been generated with the Gephi software gephi.
  • Figure 2: Networks degree distribution and the associated fit in $P(k)\sim k^{-\gamma}$
  • Figure 3: Polarization impact on epidemic dynamics: fraction of daily number of infections (right panels) and the cumulative fraction of infection (left panels)
  • Figure 4: Polarization impact on epidemic dynamics: fraction of daily number of infections (right panels) and the cumulative fraction of infection (left panels) among vaccinated individuals
  • Figure 5: Polarization impact on epidemic dynamics: fraction of daily number of infections (right panels) and the cumulative fraction of infection (left panels) among all individuals