Identifying Asymptomatic Nodes in SIS Network Epidemics using Betweenness Centrality over Time
Identifying asymptomatic individuals (i.e., infected individuals who have no clear symptoms) during epidemic outbreaks is a critical challenge, as they can transmit the disease while remaining undetected. We address this problem using a network-based susceptible–infected–susceptible (SIS) probabilistic epidemic model, where only infected and symptomatic nodes are observable at any given time instant (i.e., an epidemic snapshot). We consider the observation of multiple snapshots, with a parameter that determines the inter-observation time interval. In order to identify the asymptomatic nodes, we introduce cumulative observed betweenness (COB), an extension of observed betweenness centrality that aggregates information across multiple snapshots. When evaluated against baseline methods across diverse network models and epidemic scenarios, COB consistently achieves higher precision, which improves monotonically with the number of snapshots. We further show that the interval between observations strongly affects performance, with widely spaced snapshots providing more informative data. These results demonstrate the potential of network-based inference for identifying asymptomatic individuals under limited testing resources.