NRAND: An efficient and robust dismantling approach for infectious disease network

Abstract

Disease spread control is a challenging task with growing importance in recent years. Infectious disease networks have been proven to be a helpful resource for controlling the epidemic by targeting a smaller population. However, the information on these networks is often imprecise, diffused, concealed, and misleading, making it challenging to obtain a complete set of real-world data, i.e., some links might be missing, which can be a risk to the widespread of the pandemic. The former studies on infectious disease networks ignore the influence of neighborhood missing links in the infectious disease network topology, thus massively targeting the irrelevant population, resulting in poor epidemic control performance. In this paper, to address such a problem, we study how a small portion of the population should be targeted with incomplete network information to effectively prevent the pandemic. We propose an algorithm, namely, the Neighborhood Relation Aware Network Dismantling Algorithm (NRAND), to efficiently address the infectious disease network’s dismantling problem. For comparison, four network dismantling strategies are employed in our experiments. An extensive empirical study of real-world networks suggests that the proposed algorithm NRAND’s dismantling performance is significantly greater than the state-of-the-art algorithms, indicating that NRAND can be a smarter option for dismantling real-world infectious disease networks.