Craving-based Digital Phenotyping During MOUD Treatment

Project Summary The current opioid crisis in the United States has staggering public health implications. Medications for opioid use disorder (MOUD) reduce morbidity and mortality in OUD, however rates of return to opioid use after treatment initiation are still high. Craving is a critical component of OUD, and the heterogeneity with which it occurs may provide insight into MOUD success. Non-invasive, wearable sensors have the potential to identify objective digital biomarkers that correlate with craving, which can further be leveraged to develop digital phenotypes of OUD using data from personal digital devices. However before this technology can realize its full potential, important questions remain including the relationship between craving patterns during MOUD and treatment outcomes, and the individual level factors that best supplement craving data to identify disease phenotypes. To address this gap we will conduct a prospective, observational trial of N = 300 participants who are initiating treatment with buprenorphine, methadone or naltrexone for OUD (including of individuals whose primary drug of choice is heroin, fentanyl, or prescription opioids). We monitor participants for approximately 60 days from the time of MOUD initiation, during which continuous wearable sensor data will be passively collected. We will use digital biomarkers of craving to understand craving patterns over time, and the develop clinically informed digital phenotypes to predict MOUD responsiveness and outcomes. The Specific Aims of the proposal are to: 1) Curate a high-quality annotated dataset of craving patterns during MOUD; 2) Model trajectories of craving during initiation of treatment with MOUD and examine mediators and moderators between craving patterns and opioid use outcomes; and 3) Develop clinically informed digital phenotypes that predict MOUD outcomes. This project will provide critical insight into the use of digital biomarkers of craving the first 60 days of MOUD treatment, and into distinct phenotypes within the larger umbrella diagnosis of OUD. These data can be leveraged to modify course of MOUD therapy (dose, frequency or even dug choice), prime behavioral interventions, improve treatment efficacy, and understand contextual patterns of craving that pose threats to recovery. Furthermore, they can serve as novel measures for clinical trials to understand treatment effects in more homogeneous sub-populations of individuals with OUD. Once validated, digital phenotypes of OUD will provide opportunities to develop precision medicine approaches to MOUD treatment with just-in-time adaptive digital interventions, develop companion digital diagnostics for MOUD, and support sustained recovery.