Dynamic Treatment Regimes for Personalized Opioid Use Disorder Care - PROJECT SUMMARY This Mentored Research Scientist Development Award (K01) application is submitted by Jason Brian Gibbons, PhD, an assistant professor in the Department of Health Systems, Management, and Policy at the University of Colorado Anschutz Medical Campus. Dr. Gibbons's long-term goal is to improve the quality of care for patients with opioid use disorders (OUD) and reduce overdose deaths. Over the past few years, Dr. Gibbons has focused his research on studying the effectiveness of OUD treatment and policies and programs aimed at expanding OUD treatment service access, use, and adherence. As part of this proposal, Dr. Gibbons aims to obtain rigorous training in novel machine learning-based dynamic treatment regime methods (DTRs) and optimization trial designs to study the dynamic and heterogeneous nature of OUD severity and treatment response with the ultimate goal of using study findings to optimize OUD treatment delivery. He proposes a 5-year program of career development and mentored research to accomplish this research objective. Dr. Gibbons will work with an interdisciplinary team of mentors from the University of Colorado Anschutz Medical Campus, Brigham and Women's Hospital, and the University of Michigan, who have international reputations in the areas of his proposed training. The overarching objective of his proposal is to develop dynamic, personalized treatment recommendations using machine learning-based DTRs. DTRs can use high-dimensional data on patient's medical and social characteristics to assess the relationship between factors and treatment responses over time. The DTR then uses these relationships to craft optimal treatment decision rules. The DTRs will incorporate extensive health records from a national Behavioral Health Care Services provider, Discovery Behavioral Health, and linked all-payer claims and death records. Once the treatment decision rules have been constructed using several DTRs, a single model will be identified as “best- performing” based on its ability to generate decision rules that minimize patient risk of opioid overdose and hospitalization. A pilot hybrid factorial sequential multiple assignment randomized trial (SMART) will then be conducted to assess the acceptability and feasibility of a future full-scale trial (R01) that will evaluate the effectiveness of the generated treatment decision rules from the best-performing DTR. This research can potentially mark a paradigm shift in OUD treatment by moving towards individualized, data-driven treatment and aligns with NIDA's broader public health goals and strategic priorities. This research and Dr. Gibbons's planned training activities will prepare him to develop and test adaptive treatment strategies that may be able to improve OUD treatment outcomes. Training and research from this proposal will also be relevant to his interests in optimizing treatment for other chronic substance use and mental health disorders. Finally, the research and training will give Dr. Gibbons the solid foundation he needs to become an independent investigator and future leader in mental health and substance use disorder treatment quality research.
