Mobility and Contact Networks Shape Epidemic Outcomes: A Large-Scale Agent-Based Modeling Study
Abstract
Human mobility plays a central role in shaping contact patterns that drive infectious disease transmission, yet mobility is often simplified in agent-based models (ABMs) due to data and computational constraints. The effects of these simplifications on model outputs are poorly understood. In this study, we systematically examined how alternative mobility assumptions influence emergent contact networks and epidemic dynamics within a large-scale ABM. Using a synthetic population of one million agents representing an urban environment, we implemented five mobility models varying along two dimensions: activity patterns (empirically derived vs. randomized) and destination choice mechanisms (empirical popularity, distance-based, or random). Holding disease parameters constant, we found that mobility assumptions alone produced substantially different contact network structures and epidemic trajectories, including differences in peak incidence and shifts in outbreak timing. Importantly, these differences could not be attributed to agents simply moving more or less overall since aggregate movement volumes were broadly comparable across models. Instead, the contrasting dynamics arose from how mobility generates contact opportunities: specifically, who meets whom, where, and how often. These results suggest that in models where mobility has not been carefully calibrated, simulated epidemic outcomes and evaluations of interventions may reflect mobility assumptions as much as underlying disease parameters. Our findings underscore the importance of mobility model calibration and validation, particularly in policy-facing applications.
Related
- Synthetic populations for epidemic, demographic and mobility modeling
- A method for including additional socio-demographic factors in social contact matrices for compartment-based epidemic models
- Intermittent quarantine suppresses epidemic spreading beyond simple contact reduction
- Contagion-preserving compression for multiscale epidemic modeling
- Block-fitness modeling of the global air mobility network
BibTeX
@misc{atwal2026mobility,
title = {{Mobility and Contact Networks Shape Epidemic Outcomes: A Large-Scale Agent-Based Modeling Study}},
author = {Kuldip Singh Atwal and Emma Von Hoene and Hossein Amiri and Dieter Pfoser and Andreas Zufle and Taylor Anderson},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2607.26217},
eprint = {2607.26217},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.26217},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Mobility and Contact Networks Shape Epidemic Outcomes: A Large-Scale Agent-Based Modeling Study
AU - Kuldip Singh Atwal
AU - Emma Von Hoene
AU - Hossein Amiri
AU - Dieter Pfoser
AU - Andreas Zufle
AU - Taylor Anderson
PY - 2026
DA - 2026/10/20
PB - arXiv
DO - 10.48550/arxiv.2607.26217
UR - https://arxiv.org/abs/2607.26217
AB - Human mobility plays a central role in shaping contact patterns that drive infectious disease transmission, yet mobility is often simplified in agent-based models (ABMs) due to data and computational constraints. The effects of these simplifications on model outputs are poorly understood. In this study, we systematically examined how alternative mobility assumptions influence emergent contact networks and epidemic dynamics within a large-scale ABM. Using a synthetic population of one million agents representing an urban environment, we implemented five mobility models varying along two dimensions: activity patterns (empirically derived vs. randomized) and destination choice mechanisms (empirical popularity, distance-based, or random). Holding disease parameters constant, we found that mobility assumptions alone produced substantially different contact network structures and epidemic trajectories, including differences in peak incidence and shifts in outbreak timing. Importantly, these differences could not be attributed to agents simply moving more or less overall since aggregate movement volumes were broadly comparable across models. Instead, the contrasting dynamics arose from how mobility generates contact opportunities: specifically, who meets whom, where, and how often. These results suggest that in models where mobility has not been carefully calibrated, simulated epidemic outcomes and evaluations of interventions may reflect mobility assumptions as much as underlying disease parameters. Our findings underscore the importance of mobility model calibration and validation, particularly in policy-facing applications.
ER -