Human-Centered AI & Health Disparity | Theory-Driven Social Data Science

I am Yiwei Xu (pronunciation: ee-way shoo), incoming Assistant Professor in the College of Information at the University of Maryland, College Park (UMD iSchool) starting in Fall 2025. Prospective PhD students who are interested in working with me can apply to UMD iSchool's PhD in Information Studies program and mention my name in your application.

I study how to leverage information and technology for social good, including (a) promoting health behaviors and improving health equity, (b) understanding selective exposure and reducing polarization, and (c) addressing problematic information. I conduct experiments, surveys, and content analyses by incorporating computational methods. My work appears in venues such as Health Communication, Political Communication, International Journal of Communication, Milbank Quarterly, AAAI ICWSM, etc. I received a Top Paper Award and the Annie Lang Outstanding Dissertation Award from the International Communication Association (ICA) Information Systems Division. My research has received support through external grants from the National Science Foundation (NSF), and competitive internal grants from UW CIP, UW Population Health Initiative, UW Medicine Garvey Institute, etc.

I am currently completing my Postdoctoral Scholar training at the Center for an Informed Public (CIP) in Information School at the University of Washington, Seattle (UW iSchool). I am also a Data Science Postdoctoral Fellow affiliated with the UW eScience Institute. I received my Ph.D. from the Department of Communication at Cornell University. I have been incredibly fortunate to receive exceptional mentorship from my academic advisors: Dr. Emma S. Spiro from UW, Dr. Jeff Niederdeppe from Cornell, Dr. Erin Ash from Clemson, and I enjoy paying it forward by mentoring my advisees and students.

Research

Focused Areas

I study how to leverage information and technology for social good, including (a) promoting health behaviors and improving health equity, (b) understanding selective exposure and reducing polarization, and (c) addressing problematic information. I conduct experiments, surveys, and content analyses by incorporating computational methods.

The goal of my research is to (a) advance social scientific theory development, (b) to inform evidence-based policy, and (c) to provide implications for information and technology design.

My current research focus is on the social and psychological impacts of generative AI. There are three specific directions my collaborators and I are working on: (1) examine AI's role in health information seeking, (2) understand the role of LLMs in amplifying strategic information campaigns, (3) evaluate potential bias in LLMs' outputs and improve LLMs in delivering health information. My work is highly interdisciplinary, involving collaboration with outstanding researchers from Information Science, Communication, Computer Science, Political Science, Public Health, Psychology, etc.

Social Data Science Methods and Analytics

Traditional and Computational Experiments

e.g., Longitudinal panel experiment; Computational experiment with self-programmed mock webpage

Surveys

e.g., Survey experiment; omnibus survey

Content Analyses

e.g., TV News video content analysis with fine-grained human coding

Machine Learning

e.g., Large-scale content analysis with supervised machine learning

Big Data Analytics

e.g., Online longitudinal observational data (Google Trends) ; Text as data (Large scale news transcription extraction and analysis)

Advanced Statistical Methods

e.g., Multilevel modeling; Time series analysis; Structural equation modeling; etc.

Publications and Selected Ongoing Work

Social and Psychological Impacts of Generative AI

Xian, L., Li, L., Xu, Y., Zhang, B. Z., & Hemphill, L. (2024). Landscape of Large Language Models in global English news: Topics, sentiments, and spatiotemporal analysis. Proceedings of the International AAAI Conference on Web and Social Media (ICWSM), 18(1), 1661-1673.

Dash, S., Xu, Y., Jalbert, M., & Spiro, E. S. (in progress). Understanding the role of Large Language Models in amplifying heuristic persuasion in disinformation operations.

Liao, W., Cai, L., & Xu, Y. (in progress). Understanding people's emerging beliefs about AI as a source for seeking and processing health and scientific information.

Understanding Users' Health Information Seeking and Processing Behaviors

Xu, Y., & Margolin, D. (2024). Predictors of collective information seeking during a health crisis: A time-series and cross-sectional analysis of Google Trends during COVID-19. Health Communication. 39(2), 388-402.

Evaluating Content of Health (Disparities) in the Information Ecosystem

Xu, Y., Farkouh, E. K., Dunetz, C. A., Varanasi, S. L., Mathews, S., Gollust, S. E., Fowler, E. F., Moore, S., Lewis, N. A., Niederdeppe, J. (2023). Local TV news coverage of racial disparities in COVID-19 during the first wave of the pandemic, March-June 2020. Race and Social Problems.

Neumann, M., Moore, S., Baum, L. M., Oleinikov, P., Xu, Y., Niederdeppe, J., Gollust, S. E., & Fowler, E. F. (2024). Politicizing masks? Examining the volume and content of local news coverage of face coverings in the U.S. through the COVID-19 pandemic. Political Communication.

Xu, Y., Neumann, M., Fowler. E. F., Gollust, S. E., & Niederdeppe, J. (in progress). Community characteristics predict local news agenda building about racial health disparities.

Ash, E., Cox, E., Xu, Y., & Boatwright, B. (in progress). Promoting teen pregnancy prevention: An analysis of social media content strategy over five years.

Effects of Various Information Strategies on Promoting Health Behaviors and Evidenced-Based Policies

Xu, Y., Margolin, D., & Niederdeppe, J. (2021). Testing strategies to increase source credibility through strategic message design in the context of vaccination and vaccine hesitancy. Health Communication, 36(11), 1354-1367.

Xu, Y., Winett, L. B., Niederdeppe, J. (2021). Evidence of heterogeneity in the direction and magnitude of narrative effects on transportation and counterarguing across three independent samples. International Journal of Communication. 15. 5135–5157.

Niederdeppe, J., Winett, L.B., Xu, Y., Fowler, E. F, & Gollust, S. E. (2021). Evidence-based message strategies to increase public support for state investment in early childhood education: Results from a longitudinal panel experiment. The Milbank Quarterly. 1-44.

Winett, L. B., Niederdeppe, J., Xu, Y., Gollust, S. E., & Fowler, E. F. (2021). When “tried and true” advocacy strategies backfire: Narrative messages can undermine state legislator support for early childcare policies. The Journal of Public Interest Communications, 5(1), 45-45.

Ash, E., Xu, Y., Pool, R., Schulenberg, K., Mikkilineni, S. D., & Baraka, T. (2023). Exemplification effects on policy support: Exemplar familiarity, narrative vividness, and perceptions of maternal health disparities. Health Communication.

Xu, Y. (in progress). Communicating controversial risk issues: Effects of inoculation messages on selective exposure and subsequent persuasive outcomes.

Selected Grants

National Science Foundation (NSF) Division of Social and Economic Sciences (SBE/SES) Decision, Risk & Management Sciences (DRMS) Doctoral Dissertation Research Improvement Grant.

“Communicating Controversial Risk Issues - Effects of Inoculation Messages on Selective Exposure to Counterattitudinal Messages and Subsequent Persuasive Outcomes.” $30,168 Total. (Award #2242458)

role: co-PI; PI: Dr. Jeff Niederdeppe

February 2023 – December 2024

University of Washington Center for an Informed Public (CIP) Innovation Fund.

“Modeling the Role of Large Language Models in Amplifying Strategic (Dis)Information Campaigns and Examining its Persuasive Effects.” $23,465.35 Total.

role: PI; co-PI: Saloni Dash

March 2024 – June 2025