Fig. 5.

The effectiveness of clustering methods in finding high-risk individuals. The average number of new infections between years 9 and 10 of the simulation caused by individuals infected at year 9 in growing clusters. We select 1000 individuals from clusters, inferred by either HIV-TRACE or TreeCluster, that have the highest growth rate (ties broken randomly). As a baseline control, the average number of infections over all individuals (similar to expectations under a random selection) is shown as well. For a cluster with n_t members at year t, growth rate is defined as $\frac{n_9-n_8}{\sqrt{n_9}}$. The columns show varying expected degree (i.e. number of sexual partners), and all other parameters are their base values