PROJECT ABSTRACT Chicago is a national epicenter of opioid overdose (OD) and related harms. Opioid-related Emergency Service Calls (ORESCs) are critical opportunities for service engagement and intervention. Nationally and in Chicago, high mortality rates subsequent to non-fatal OD underscore that such opportunities are often missed. Chicago will pilot two innovative responses to address these challenges. Alternate Immediate Response (AIR) will provide assertive outreach and engagement, including connecting people with medication opioid use disorder treatment and other related services. Alternate Immediate Response plus Follow-up (AIR-F) includes AIR, along with 8 weeks of follow-up services. AIR response teams consisting of a community paramedic and a peer support specialist/recovery coach will be deployed to provide assertive outreach services, including developing treatment, safety, and follow-up plans, delivering brief interventions, and providing transport to pertinent other services. AIR-F engagement will include linkages to follow-up services that promote treatment retention, including case management, care coordination, and connections to community-based care and treatment. We propose to use machine learning (ML) to develop a tool to identify individuals at highest risk of OD and OD-related mortality in order to prioritize service delivery and follow-up services. In particular, we will develop a random forest (RF) classifier that will combine data from the Chicago Department of Public Health, the Office of Emergency Management and Communications, Chicago Police and Fire departments, Chicago Office of Public Safety Administration, and Cook County Medical Examiner’s Office to create an integrated dataset to trace emergency calls from origination to final disposition. We will also extract data from unstructured text included in CFD ambulance data. By incorporating multiple large administrative datasets, the tool will capitalize on diverse sources of “signal,” maximizing prediction accuracy. We will then use our integrated data to predict individuals at highest risk of subsequent ORESCs, OD, arrest, and other adverse outcomes. AIR/AIR-F staff will use these indicators along with other clinical data to allocate scarce follow-up resources. We will use difference-in-differences estimation to compare post-intervention outcomes within the service area on Chicago’s west side (Humboldt Park, West Garfield Park, and East Garfield Park) to that observed in contiguous communities (Austin, North Lawndale, South Lawndale, Lower West Side, and West Town) to gauge the population impact of AIR/AIR-F. Finally, we will conduct qualitative interviews with various stakeholders to provide additional insight into the pilot and explore how Chicago can better serve at-risk individuals. To help reduce the prevalence of ORESCs’ and associated mortality, all computer code developed for this grant will be made available open-source. We will disseminate project findings...