A Co-Creation Framework for Eliciting Human Behavior Models to Support Crisis Risk Management: Evidence from Greece

Abstract

Human behavior models (HBMs) are representations of how individuals perceive risk, make decisions, and act during crises. These models are increasingly used in computational simulations to support disaster risk management, e.g., in assessing current or devising future evacuation strategies. However, current HBMs are often developed in an ad hoc manner and remain weakly grounded in empirical and participatory evidence. This limits the realism and transferability of simulations, ultimately reducing their value for crisis risk management. Accordingly, this paper presents a generalizable, iterative co-creation framework for eliciting HBMs. The framework translates contextual, social, and decision-making knowledge into validated, model-ready behavioral representations through empirical data, expert validation, and continuous stakeholder feedback. To evaluate the framework, it was applied in Egaleo, Greece, for earthquake and heatwave scenarios, where evacuation strategies were tested using agent-based simulations coupled with elicited HBMs. The results of simulations reveal significant differences in evacuation timing and spatial congestion.