Epidemic models have long been used to study information diffusion in complex artificial, biological and social networks. Here, we propose a novel recurrent-state epidemic model for information diffusion in networked systems by introducing a potential-spreaders compartment into the susceptible–infected–recovered–susceptible (SIRS) model. Our model assumes that not all susceptible nodes are equally likely to become efficient information spreaders, and that information can be repeatedly disseminated in cycles, even after its temporary decay. We observed that the introduced potential spreaders compartment in our model enables a more convenient state-transition process and an accurate description of information diffusion based on discrete time. Specifically, we found that our analytic results are in good agreement with numerical simulations on both artificial systems and eight different types of real-world biological, medical, and social networks. Unlike susceptible or recovered nodes, we further observed that potential spreader nodes in our model can serve as a relatively good predictor of the peaks of contagion outbreaks. Monitoring the number of potential spreaders could thus be beneficial for predicting both infection transmission and information propagation in complex networks and for characterizing their temporal dynamical patterns. Our model could be applied to a wide variety of spreading phenomena with recurrent-state endemic dynamics, such as seasonal influenza, re-tweeting behavior, or re-emergence of rumors and fashion trends in online social networks.

流行模型长期以来一直用于研究复杂的人工，生物和社会网络中的信息传播。在这里，我们通过将潜在的传播者隔室引入易感，感染，恢复，易感（SIRS）模型中，为网络系统中的信息扩散提出了一种新的复发状态流行病模型。我们的模型假设并非所有易受感染的节点都同样有可能成为有效的信息传播者，并且即使在信息暂时衰减之后，信息也可以周期性地重复传播。我们观察到，在我们的模型中引入的潜在扩展器隔室使状态转换过程更加方便，并基于离散时间对信息扩散进行了准确的描述。具体来说，我们发现，我们的分析结果与人工系统和八种不同类型的现实世界生物，医学和社会网络上的数值模拟非常吻合。与易感节点或恢复节点不同，我们进一步观察到，模型中的潜在扩展节点可以作为传染病爆发高峰的相对较好的预测指标。因此，监视潜在散布器的数量可能有助于预测复杂网络中的感染传播和信息传播，并有助于表征其时间动态模式。我们的模型可以应用于具有周期性流行病的各种传播现象，例如季节性流感，重新发布行为或在线社交网络中谣言和时尚趋势的重新出现。