Adaptive resource sharing strategies for grassroots epidemic response: framework, inference and evaluation

Effective epidemic response relies on well-designed emergency resource sharing among decentralized grassroots actors operating under highly uncertain propagation dynamics. However, insights into how such sharing can be structured and coordinated remains insufficiently understood. This study investigates emergency resource sharing at the grassroots level of epidemic response, addressing uncertainty in heterogeneous resource demands arising from dynamically evolving epidemic scenarios. To systematically represent the complexity of scenarios faced by grassroots decision-makers, we develop a three-dimensional scenario model that characterizes variations in epidemic outcomes. Meanwhile, we propose a resource sharing framework grounded in “actor-task-resource” relationships, which captures the scenario-adaptive and task-driven resource demands of grassroots actors. To infer coordinated resource sharing decisions under scenario uncertainty which minimise the expected total shortage and transport costs, we employ a distributionally robust optimization (DRO) approach which incorporates time-varying scenarios through switching Wasserstein ambiguity sets. The model is reformulated into a tractable mixed-integer linear program (MILP) via duality theory. Numerical experiments demonstrate that the proposed resource sharing strategies effectively mitigate resource shortage risks for grassroots units across uncertain scenario realizations. Moreover, combining external resource support with internal resource sharing yields superior performance, especially when this prioritizes high-risk areas. Acquiring limited external support across small amounts of multiple resource types instead of restricting to a single resource is also shown to reduce shortage risks given uncertain scenarios. These findings offer practical insights for designing resilient grassroots emergency resource coordination plans that adapt to evolving epidemic conditions. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2026.