Agent-based models have proven to be useful tools in supporting decision-making processes in different application domains. The advent of modern computers and supercomputers has enabled these bottom-up approaches to realistically model human mobility and contact behavior. Taking the next step and combining realistic agent-based models with up-to-date or real-time data yields powerful agent-based digital twins that can help to save lifes or protect human health endangered in situations of disaster or catastrophes. In [1], the authors have recently presented a city-scale agent-based model that was capable to evaluate advanced testing and isolation strategies during the COVID-19 pandemic. Bi-directional coupling of simulations as proposed in [2], at the example of an agent-based pedestrian simulation with an electrical simulation in an airport terminal, allows to test contingency plans in saboteur scenarios. However, for agent-based digital twins to be useful in different situations of disaster very different requirements need to be fulfilled. For large-scale catastrophes such as pandemics or country-wide network shutdowns, the corresponding agent-based models need to scale well on supercomputing infrastructure to allow the simulation of millions of agents in a fraction of second. For small-scale perils such as the local leakage of a chemical hazard, fine-granular pedestrian dynamics are needed. Moreover, future integration of real-time data with agent-based simulations could leverage its applications in adaptive emergency planning. By attending this symposium, participants will gain a deeper understanding of agent-based digital twins that can help decision makers and domain experts to protect humans from danger caused by disasters of different nature. The symposium will explore different approaches to agent-based digital twins with its domain specific challenges, including: • Fine-granular pedestrian movement models to protect humans in situations of chemical, thermic or mechanical dangers or catastrophes • Small- to large-scale models to react to network or cyber catastrophes • Large-scale models to protect human health in times of pandemics • Impact estimation of disasters in the context of contingency planning and risk analysis • Integration of real-time data with agent-based models for improved emergency management