Yuke Wang
Bio
My research interests span the areas of pathogen shedding kinetics, wastewater surveillance, exposure assessment, and infectious disease transmission. A common thread in my research is in understanding the human-environment interface with respect to infectious diseases. I design, develop, and evaluate statistical and mathematical models to support evidence-informed public health decision-making in low- and middle-income countries (LMICs) and the United States. These model frameworks were packaged as open-source tools and deployed across the globe.
More recently, that work has extended to artificial intelligence. I build AI-powered systems for evidence synthesis (e.g., multi-agent pipelines that discover, extract, and review biomarker shedding data) and AI agents that conduct epidemiological modeling through plain text conversations. The motivation is the same as the rest of my work: the bottleneck in public health is rarely the mathematics, but the effort of assembling trustworthy evidence and the expertise needed to use it.
Over the last decade, I have been involved in projects funded by NIH, the Bill & Melinda Gates Foundation, Wellcome Trust, and the Rockefeller Foundation, on the aforementioned research areas for various infectious diseases. Half of these projects were in LMICs, where I have developed strong relationships with local partners. I enjoy working on team science (cross-disciplinary collaborations from diverse scientific fields) and contribute my analytical and AI skills as a toolbox for real-world problems.
Areas of Interest
- Artificial Intelligence
- Data Science
- Bayesian Analysis
- Biostatistics
- Disease Surveillance
- Exposure Assessment
- Global Health
- Infectious Disease Dynamics
- Modeling
Education
- PhD, Georgia State University
- MSPH, Emory University
- Bachelor, South China University of Technology
Affiliations
Current Activities:
Shedding Hub: The Shedding Hub collates data and statistical models for pathogen shedding in different human specimens, such as stool or sputum samples. Developing wastewater-based epidemiology into a quantitative, reliable epidemiological monitoring tool motivates the project. It now holds curated measurements from 100+ studies across 16+ pathogens, with fitted shedding curves and a Python package so a transmission or wastewater model can consume the parameters directly. Studies are curated by a multi-agent AI pipeline in which extraction and review run on different model providers, and a human curator in charge of final review and quality control.
EpiChat: EpiChat turns a plain-language question into a validated epidemiological simulation. Demand for modeling has long outpaced the supply of trained modelers, most sharply during COVID-19 and in LMICs, and EpiChat is an attempt to remove that bottleneck: a policy officer, surveillance team, or health ministry analyst can explore an epidemic scenario in minutes rather than waiting weeks for specialist support. It supports six compartmental model structures and draws on demographic and health data spanning 237 countries.
Where Do We Poop?: Wastewater-based epidemiology usually assumes people excrete at home and that populations sit still. Neither is true, and both assumptions distort what a sewer sample means. We built agent-based geospatial simulations that combines realistic urban activity patterns, a physiologically motivated defecation cycle, disease transmission, and pathogen shedding. Here is an example of our work: PoopSimCity.
In The News
