Hybrid metapopulation-agent-based epidemiological models for efficient insight on the individual scale: a contribution to green computing

Kurzfassung

Over the past decades, mathematical models have contributed in understanding dynamics of infectious diseases and are of great aid when it comes to finding suitable intervention measures. However, fine granular data might be needed for realistic scenarios. In addition, they may need substantial computational effort producing significant CO2 emissions. Two popular modeling approaches for the simulation of infectious disease dynamics are agent-based and population-based models. Agent-based models (ABMs) offer a microscopic view and are thus able to capture heterogeneous human contact behavior and mobility patterns. However, insights on individual-level dynamics come with high computational effort that scales with the number of agents. On the other hand, population-based models (PBMs) e.g. based on ordinary differential equations (ODEs) are computationally efficient even for large populations as their complexity is independent of the population size. Yet, population-based models assume a (to some extent) homogeneous and well-mixed population and are therefore restricted in their granularity. To manage the trade-off between computational complexity and level of detail, we propose spatial- and temporal hybrid models that use ABMs only in an area or time frame of interest. For the spatial hybridization, we use population-based models to account for relevant influences to disease dynamics from outside, e.g. due to commuting activities. For the temporal hybridization, we use population-based models when single simulation results of the agent-based model come closer to averaged outcomes. We investigated simulation outcomes of the proposed hybrid models for particular scenarios as well as their scaling behavior, demonstrating significant reduction in computational effort by up to 98% - without losing the required depth in information in the focus frame. Furthermore, we analyzed how parameters impact the model runtime, finding that the runtime of the ABM depends not only on the number of agents in the system, but also on the number of transmissions. In contrast, the runtime of the population-based model remains unaffected by the number of transmissions. Eventually, we compared the CO2 emissions across all models for the considered scenarios, showing a saving of 70-90% for the hybrid models.