Immunization strategies

When a new disease emerge (e.g., COVID 19 pandemic), the vaccine production takes time and at the early stage only a small number of doses are available. In this situation whom should we immunize first?

Network-based insights

From the Network science perspective, Hubs are the most important nodes to immunize because epidemic spreading is driven by the hubs when there is a strong heterogeneity in degree. It is closely related to Percolation and the question of how to efficiently break the network down. In a Scale free network, random immunization doesn’t work efficiently; only degree-preferential immunization can block the disease effectively with a small fraction of immunization.

Another consideration is Community structure and Betweenness. Immunizing connectors can be highly effective at blocking the transmission across communities, containing the disease. Thus many strategies use Betweenness or similar quantities (e.g., local clustering coefficient in Xia2017improved). See Salathé2010dynamics.

One neat trick is Acquaintance sampling, suggested in Cohen2003efficient and Christakis2010social. Essentially, if we sample a random person and immunize their acquaintances (not the sampled person), we can effectively perform immunization proportional to their degree. However, note that the sampling happens from the friendship (acquaintance) network, which may be distinct from the network through which the disease is spreading. Also in practice, it is essential to consider many other factors (e.g., healthcare workers vs. others, age-based fatality, social equity) and thus it is difficult to implement this acquaintance immunization.