SCC-IRG Track 1: A New AI-driven Paradigm to Promote Community Resilience for Teenagers and Young Adults in Preventing Opioid Misuse and Addiction

Combating the deadly opioid epidemic is a national priority. Specifically, teenagers and young adults (TYAs) are disproportionately affected by and particularly vulnerable to opioid misuse and addiction. Unfortunately, research on how to provide effective yet affordable solutions to generate and promote tailored messages for the distinct yet vulnerable TYA community against the opioid crisis is lacking.

To fill this gap, the “smart and connected (S&CC) community” in this project is defined as the community of at-risk TYAs who are particularly vulnerable to opioids and connected via online social media, within which intelligent technologies will be synergistically integrated to improve their resilience and well-being against the lethal opioid epidemic. By engaging with representative community stakeholders, this project will design and develop a new AI-driven paradigm to facilitate personalized messages tailored to TYA community’s characteristics and circumstances to promote their resilience against opioid misuse and addiction, and thus help enhance national public health, safety, and welfare.

The developed framework can be scaled and easily transferred to other communities in preventing and reducing corresponding harms, such as people with different types of substance misuse and adolescents at-risk for suicide. The proposed work will advance scientific theory in related research communities and benefit multidisciplinary domains, including epidemiology, economics, and social and behavioral sciences.

This research will accelerate personalized interventions for the at-risk TYA community in reducing opioid misuses and overdose deaths. First, based on the large-scale data generated by TYAs on social media, the team will develop graph neural networks with novel self-supervised and prompt learning techniques to detect at-risk TYAs, by addressing the challenges of heterogeneity, multi-modality, and limited labeling of the online data.

Second, to derive and interpret key factors from diverse contexts presented by at-risk TYAs, the team will develop a novel causal analysis framework by bridging knowledge graph and large language model for intricate reasoning. Third, given the detected at-risk TYAs with related causal analyses, the team will develop a safety-enhanced multi-modal learning framework aiming to generate messages tailored to at-risk TYAs in preventing opioid misuse and addiction.

Fourth, the team will further develop an adaptive reinforcement learning framework enabling user-in-the-loop to facilitate an interactive process that could inform users’ feedback, calibrate the generated messages, and thus produce adaptive interventions for at-risk TYAs. This project will integrate research with education and the outcomes will be made publicly accessible and broadly distributed.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.