Best Student Poster at GeoMed Conference 2026
At the recent "GeoMed" conference 2026 held in Pamplona, Spain, the prize for the best student poster was awarded to Poshan Niraula for his poster titled "Spatiotemporal Causal Analysis of Human Mobility on Disease Transmission". The work was co-authored with Jorge Mateu and Edzer Pebesma.
The work combines methods from spatio-temporal modelling with epidemiological data on COVID-19, in an effort to determine the effect mobility changes may have on COVID-19 transmission.
Summary:
The study aimed to quantify the causal effect of human mobility on the weekly COVID-19 case rates across the 177 Modified Zip Code Tabulation Areas (MODZCTA) of New York City. We deployed generalized propensity score (GPS) framework [1] with spatial interference using Advan origin–destination mobility dataset [2] to estimate:
- direct effect: the effect of mobility change within a focal MODZCTA and
- spillover effect: the effect of mobility change in other MODZCTAs, weighted by the volume of movement between them.
Across all the fitted models, the spillover effect was consistently more pronounced and statistically significant compared to the direct effects. In several instances the direct effect was also negative, which we interpret as residual time-dependent confounding, rather than protective effect of mobility. We aim to improve on this research by relaxing the unmeasured confounding assumption, and explicitly accounting for the time-dependent confounding.
[1] Giffin, A., Reich, B. J., Yang, S., \& Rappold, A. G. (2023). Generalized propensity score approach to causal inference with spatial interference. Biometrics, 79(3), 2220-2231.
[2] Advan Research. Foot traffic / weekly patterns plus [dataset], 2025. DOI: https://doi.org/10.82551/C103-N851.