Multi-Medical Resource Collaborative Allocation for Emerging Epidemics Based on Trend Learning-Driven Prediction

The containment of emerging infectious disease outbreaks presents formidable challenges for cross-regional medical resource allocation, due to the inherent characteristics of unforeseen outbreaks, rapid spread, and spatiotemporal uncertainty in disease progression. The common practice in solving this problem is first predicting the number of infections and then utilizing these predictions as inputs for resource allocation. However, this decoupled approach fails to account for the critical interdependency between resource allocation decisions and disease transmission dynamics, where resource allocation directly influences infection trajectories, while simultaneously evolving case numbers alter resource requirements, which leads to prediction inaccuracies and inefficient resource allocation. To address these challenges, this study establishes an integrated framework that combines a modified epidemic prediction model with cross-regional transmission and trend learning. Building on this dynamic characterization, we formulate a mathematical programming model to minimize population exposure through coordinated allocation of heterogeneous medical resources. Due to the highly nonlinear coupling relationship between the resource allocation decision and transmission dynamics, it is intractable to solve the model exactly. To address the computational complexity, we develop a variable neighborhood search algorithm that exploits problem-specific structures to construct neighborhoods. Extensive numerical experiments using real data of the COVID-19 epidemic demonstrate the effectiveness of our proposed prediction model and the superiority of our proposed resource allocation approach compared with benchmark policies.