Sudden shifts in vaccine uptake, vaccine opinion, and infection incidence can occur in coupled behaviour-disease systems going through a bifurcation as the perceived risk of the vaccine increases. Literature shows that such regime shifts are sometimes foreshadowed by early warning signals (EWS). We propose and compare the performance of various measures of network structure as potential EWS indicators of epidemics and changes in population vaccine opinion. We construct a multiplex model coupling transmission of a vaccine-preventable childhood infectious disease and social dynamics concerning vaccine opinion. We find that the modularity of pro- and anti-vaccine network communities perform well as EWS, as do several measures of the number and size of opinion-based communities, and the size of pro-vaccine echo chambers. The number of opinion changes also gives early warnings, although the clustering coefficient and metrics concerning anti-vaccine echo chambers provide little warning. Stronger social norms are found to compromise the ability of all EWS metrics to provide advance warning. These exploratory results suggest that EWS indicators based on the network structure of online social media communities might assist public health preparedness by providing early warning of potential regime shifts.

随着对疫苗的感知风险的增加，在经历分叉的行为-疾病耦合系统中，可能会发生疫苗接种、疫苗意见和感染发生率的突然变化。文献表明，早期预警信号 (EWS) 有时预示着这种政权转变。我们提出并比较了网络结构的各种措施的性能，作为流行病和人口疫苗意见变化的潜在 EWS 指标。我们构建了一个多重模型，将疫苗可预防的儿童传染病的传播与有关疫苗意见的社会动态耦合起来。我们发现，支持和反对疫苗网络社区的模块化表现与 EWS 一样好，对基于意见的社区的数量和大小以及支持疫苗回声室的大小的几种测量也是如此。意见变化的数量也提供了预警，尽管有关反疫苗回声室的聚类系数和指标几乎没有提供任何警告。发现更强的社会规范会损害所有 EWS 指标提供预警的能力。这些探索性结果表明，基于在线社交媒体社区网络结构的 EWS 指标可能通过提供潜在政权更替的早期预警来帮助公共卫生准备。