University of Rochester School of Medicine Rochester, New York, United States
Abstract Body This study uses agent-based modeling (ABM) to examine how external social network processes moderate individual therapy outcomes for people receiving individual psychotherapy. Specifically, we simulate a population of individuals with varying levels of social influence on one another (i.e., contagion), preferential attachment to others with similar risk factors (i.e., homophily), and vulnerability to background risk factors (e.g., shared risk factors from the broader environment, analogous to the general health of local workplace or school). We evaluate the initial validity of this evolving social network model by showing it reproduces well-established patterns of suicide risk in the empirical literature: increased risk of suicidal behavior at the periphery of a social network, formation of persistent suicide clusters, rapid onset and remediation of suicide risk for individuals in those clusters. With validity of the underlying model established, we then apply interventions at the individual level (e.g., analogous to traditional psychotherapy) and show that a person’s responsiveness to such treatment is powerfully moderated by the external dynamics of the social network in which they are embedded – factors currently outside the control of standard psychotherapy techniques. Lastly, we show that a new class of Network Enhancing Interventions (NEIs) – which seek to augment the social integration and behavioral norms within a pre-existing social group – are not only more durable against these network effects than individual treatments, but in fact benefit from them. Discussion focuses on building a broader suicide prevention portfolio that includes both traditional individual interventions (e.g., psychotherapy), as well as techniques to directly target the social networks around the people receiving those individual interventions. This also includes successful examples of clinical trials employing NEIs in the empirical literature, highlighting their feasibility for broader application.