In closed rooms with limited convection human motion can considerably affect the airflow and thus the dispersion of pollutant. However, in Computational Fluid Dynamics (CFD) simulations on air quality and safety for human beings this effect is generally not considered, which is mainly due to a lack of a well-founded and detailed estimation of the human behavior and the high computational cost of taking into account moving objects in CFD meshes. This work addresses this issue by coupling multidisciplinary methods to allow for a more realistic simulation of pollutant dispersion by taking into account the influence of human movements. A Social Force Model predicts trajectory and speed of each person moving in a complex environment. A lattice Boltzmann-based CFD tool provides a Large Eddy Simulation of the unsteady turbulent airflow with pollutant dispersion and thermal effects. And an Actuator Line Model supplies the CFD tool with body forces that mimic the impact of moving objects on the airflow, thus, avoiding computationally expensive dynamic meshing. The capability of the coupled model is demonstrated on three realistic evacuation scenarios with various pollutant sources and a wide range of scales (dimension from 10 to 100 m, occupation from 10 to 6000 persons). The results allow to access instantaneous environmental parameters like pollutant concentration for each person during the course of the evacuation, enabling the assessment of associated health risks.