Optimizing a Digital Support Service for Concerned Significant Others of Individuals with SUD with Helpline, AI-Enhanced and Peer-Led Support - Project Abstract Families of individuals with substance use disorder (SUD) experience high levels of distress and limited access to evidence-based support. Concerned Significant Others (CSOs), particularly parents and caregivers, often struggle to navigate their loved one's (LO) substance use and treatment options due to stigma, financial barriers, and systemic exclusion from care. While family involvement improves treatment engagement and recovery outcomes, few scalable interventions exist to provide structured, ongoing support to CSOs. Digital interventions, including AI-driven chatbots, have the potential to increase accessibility and engagement; however, no AI-based models are currently designed to deliver tailored, evidence-based support to CSOs. This R61/R33 study will develop and evaluate HALO (Helping Affected Loved Ones), an AI-powered chatbot designed to provide real-time, structured guidance to CSOs within a digital family support ecosystem. HALO will be integrated into the Partnership to End Addiction's Helpline, which serves over 23,000 families annually through phone, text, automated messaging, peer support, and e-learning. In the R61 phase, we will apply natural language processing (NLP) techniques to analyze 40,000+ Helpline interactions, conduct surveys and interviews with CSOs, LOs, and peer coaches, and use a user-centered design approach to develop HALO. A three-month pilot trial (N=60) will assess feasibility, usability, engagement, and preliminary effects on family recovery capital, CSO functioning, and LO substance use across three conditions: (1) Helpline (HL) only; (2) HL + HALO; and (3) HL + HALO + peer support (PS). Meeting predefined feasibility benchmarks will determine progression to the R33 phase, where we will conduct a six-month, 2x2 factorial trial (N=800) to evaluate the individual and combined effects of Helpline, HALO, and PS on CSO and LO outcomes. A sub-sample of 200 LOs will be assessed at three months to examine the experiences and perceptions of LOs and their relationship with CSOs on their recovery. This study will provide the first rigorously tested AI-human hybrid support model for CSOs and generate data on the comparative and combined effects of AI-driven and peer- based support interventions in a representative group of CSOs with different needs. Findings will expand real- world tools for CSOs by integrating HALO into the Partnership's free digital support ecosystem and making it available to treatment and community-based services that lack formal family support interventions.
