The Advanced Methods and Research Core (AMRC) will provide a suite of rigorous capabilities to enhance the validity, generalizability, and policy impact of JCOIN clinical trials. Network and spatial analyses are rarely included in traditional trial designs. These tools allow researchers and clinical interventionists to explore questions such as the consequences of transitions from prescription opioids to street drug use. Predictive Analytics helps researchers to predict, usually through machine learning, which opioid users are at highest risk of re-offending, fatal or non-fatal overdose, and thus who should be provided the most-intensive outreach and service resources. Agent-based modeling (ABM) similarly helps researchers to explore underlying mechanisms and epidemiological processes such as the health and mortality pathways of individuals who experience non- fatal overdose (OD) or who initiate medication-assisted treatment (MAT). By illuminating mechanisms associated with intervention success, these advanced modeling approaches allow exploration of additional unobserved or downstream outcomes in specific trials. These methods help to identify specific intervention ingredients and mechanisms likely to have the most impact on overdose elimination in the US. The AMRC will utilize the opioid Cascade of Care (CoC) framework to pursue three advanced analytic projects that model multiple outcomes within diverse justice contexts. The modeling projects are sequenced to move along the Cascade to model specific outcomes including opioid use disorder identification (project #1); diagnosis of opioid use disorder (project #2); and interventions to impact MAT engagement and retention, and other downstream outcomes (project #3). The AMRC has a history of advancing these models to: create rigor through component integration, disciplined model validation, and transparent processes to ensure stakeholder and community perspectives provide input on key modeling and parameter assumptions, and resulting outputs. JCOIN clinical trials will only be able to answer some complicated, though trial data can be incorporated and used for calibration in complex models. Advanced models provide helpful guidance when interventions influence complex behaviors that are not directly measurable (e.g. # needle-sharing partners or decisions to inject alone), when trials are not feasible or ethical, and when interventions influence multiple downstream outcomes (e.g. HIV and HCV) likely to be subject of multiple and interacting interventions. The AMRC will employ three modeling approaches that collectively address these challenges. In particular, the AMRC aims are to: 1) Determine combined social and spatial network drivers of opioid use disorder; 2) Develop predictive models of OD within JCOIN justice contexts; 3) Extend and maintain an Agent-based Model framework of JCOIN opioid use interventions to enhance clinical trial capacity including cost-effectiveness of tested interventions.
