Reproduce “From Urban Mobility to Epidemic Dynamics: A Mixture-of-Experts Framework with Preference Alignment for Policy Scenario Simulation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yun Ye, Arsalan Dezhkam, Junyuan Liu, Xinglei Wang, Tao Cheng
A data-driven framework, UrbanShare-MoE-PA, developed at University College London, simulates the behavioral, epidemiological, and urban activity impacts of non-pharmaceutical intervention (NPI) calendars by generating agent-level daily mobility patterns. The framework demonstrated that early lockdown timing was more effective than extended duration in reducing epidemic burden, while revealing trade-offs with urban activity levels.
Key insights
- Modeling heterogeneous behavioral responses through a sparse Mixture-of-Experts architecture significantly improves the fidelity of predicting individual activity and mobility patterns under NPIs.
- Explicitly aligning behavioral predictions with policy phases via preference learning and distribution-shift augmentation is critical for generating temporally coherent and plausible alternative-calendar scenarios.
- The timing of NPI implementation, particularly before the epidemic peak, can have a greater impact on reducing epidemic burden than simply extending the duration of restrictions, highlighting non-linear trade-offs between public health and urban activity.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.