Hsuan-Wei “Wayne” Lee, assistant professor, department of biostatistics and health data science at Lehigh University’s College of Health (COH), has received an AI Integrated Grant from the Office of the Vice Provost for Research (OVPR) to create a novel tool to assess opioid addiction risk.
The project, titled AI-Powered Opioid Addiction Risk Prediction via Social Network Dynamics, aims to shift the focus from late-stage detection of opioid misuse to an early identification of risk. By analyzing how risk spreads through social networks and digital environments, this research seeks to enable more proactive and targeted public health interventions. Addressing real-world public health challenges that affect our local, national and global communities is a core tenet of the COH.
Lee’s work centers on building a mathematical model grounded in evolutionary game theory to better understand the behavioral drivers of substance use, such as peer pressure. He will also examine broader socio-economic factors, including crime rates and unemployment, that impact opioid usage. For this project, Lee will collaborate with Mooi Choo Chuah, professor of computer science and engineering in the P.C. Rossin College of Engineering and Applied Science, to develop the system. Her expertise complements Lee’s own expertise in network science, applied mathematics and epidemiology. Likai Wang ‘25, who is pursuing a MS in data science, will help to build the system.
Together, they will leverage graph neural networks, a form of AI that learns from relationships and connections within data, to model how opioid use propagates across networks. By inputting large-scale real data such as CDC mortality data, Medicare prescribing patterns and census indicators, they will create a large system that can simulate opioid risk patterns. The model will integrate multiple layers of information to generate a more comprehensive view of how opioid risk emerges and spreads.
“We can make a calculation in the system,” Lee explained. “We can have different layers of the networks because all of the hospitals, individuals and counties are connected. If some of my close friends are using drugs, then I have a higher chance to use the drug. In the hospital, if there were many substance abuse cases, then neighboring hospitals will probably have similar circumstances.”
The model will allow the researchers to simulate different scenarios by adjusting factors such as addiction and relapse rates. “It’s like predicting the weather,” he said. “You can have a very complicated model, and then you can have a group of people connecting the data from the real world. You adjust your model in real time.”
The study, which began in February 2026, is expected to conclude in the late summer. Lee hopes to use the pilot results to pursue a larger, multi-year grant from the National Institutes of Health (NIH), while also creating research opportunities for COH students.
“I’m really thankful Lehigh gave us the chance because we need to start from something small,” Lee said.
Currently, Lee is also leading a Creative Inquiry team, which includes several COH students, in a NextGen Impact Fellowship called Behavior-Adaptive Epidemic Digital Twin for Community Health Resilience. The team will be studying the rationale behind substance abuse, and using this insight, they will build a machinery called a digital twin to simulate how virtual populations adapt and react to public health interventions and information campaigns. In the future, he hopes to merge the AI-Powered Opioid Integrated Addiction Risk Prediction project with this one, given the synergies between the two.
Lee has also partnered with Joseph A. Pacheco, assistant professor, department of community and global health and director, Institute for Indigenous Studies; and Won Choi, professor, department of population health and associate dean, research and graduate studies, on related work which analyzes addictive behavior in Native American communities and the spread of substance use through networks. Lee’s work on mathematical modeling of addictive behaviors was published earlier this year in ScienceDirect, and he has also partnered with U.C. Berkeley on a research project examining how eliminating California’s cultivation tax in 2022 has reshaped retail sales, market structure and public health implications.
Together, Lee’s efforts reflect a broader goal to move from reactive responses to substance use toward more predictive, prevention-focused strategies that utilize innovative technologies to improve public health outcomes.
