Source Identification of Asymptomatic Spread on Networks

Identifying the sources of spreading dynamics on networks has drawn extensive attention in recent years, where a variety of epidemic models have been adopted to simulate the propagation dynamics among network nodes, such as the Susceptible-Infectious (SI) model. The objective is to identify the most possible source of infection based on the observed state transition events (e.g., from susceptible to infectious) on a small set of monitored nodes. Most existing studies assumed that once a monitored node is infected, it will be immediately observed. However, in reality, it is likely that for many infectious diseases, a newly infected person becomes infectious without showing any disease symptoms. In this case, the state transition cannot be observed until symptoms arise after a time period (i.e., the incubation period). Accordingly, in this paper, we focus on investigating the source identification problem of asymptomatic spread on complex networks by monitoring only a small number of network nodes. Specifically, we adopt a continuous-time SI model with a contagious incubation period to simulate the asymptomatic spread on networks, where the length of the incubation period is assumed to be an independent and identically distributed exponential distribution. In doing so, we formulate the source identification problem as a likelihood maximization problem and solve it using the Monte Carlo approximation and importance sampling. Finally, we validate the performance of our method on both synthetic and real-world networks with different experimental settings. The results show that our method can achieve higher identification accuracy than several benchmark methods.