Transportation networks have a major influence on the interactions between people. The worldwide diffusion of some disease, like HIV, SARS, H1N1 or MRSA has for instance been eased by the recent increases in the velocity of transportation means and the frequency of movements. In this context, it is of crucial importance to develop theoretical tools to understand the influence of human mobility on infectious disease propagation. During this thesis, we have performed at first a systematic review of the use of population movements in models of epidemic propagation. Then, we focused on the national/regional scale, using commuting movements, and we developed several tools to analyze their influence on the propagation of influenza in France. The first part of this thesis has been dedicated to the systematic review of works on epidemic propagation integrating mobility data. Nowadays, the influence of air travel on disease diffusion has been extensively studied: on national/regional scale however, commuting movements are often integrated in models without justification. In the following of this thesis, we investigated the influence of these movements: the example of influenza propagation in France was used to make this study. We started this work with the calculation of spatial auto-correlation of surveillance data from the Sentinelles network, based on commuting data, in order to confirm they had an influence on influenza propagation. Then, we developed tools based on a metapopulation model, to compare epidemics propagation, thanks to which we showed that both children and adults movements, which are different, should be integrated in models. In this study, we also evidenced the existence of preferential pathways for diffusion, that are followed depending on the seeding district of the epidemic. We then performed an analytical analysis of commuting movements influence by developing a linear model based on the previous one. In this thesis, we propose two tools based on the spectral analysis of the matrix of this model and the study of the vectorial sub-space generated by its dominant eigenvectors: with these tools, we studied the role of each district in the propagation on one hand and the global dynamics of the model on the other hand. Previous results were confirmed by this second analysis.
