Quantifying the impact of school holidays and week-ends on the spread of influenza-like-illness: a multi-scale approach (R-6638)

It is known that children fuel the onset of a flu epidemic, and that their social and mobility patterns vary significantly between school term and periods of school closure. School holidays may therefore have a considerable impact on the spread of a flu epidemic. The quantification of such phenomenon would provide a more comprehensive understanding of the dynamics of flu spread in a population and also offer important information and guidelines for public health management, in terms of for example control policies such as school closures and selective vaccination. Many works have already investigated the topic of school closure either at very coarse-grained scale, without accounting for spatial structure [1,2] or, when space structure was taken into account, using a very stylised model [3]: until now, none have attempted to address this issue under more realistic assumptions. During this collaboration we aim to develop a high-resolution, data-driven spatial model for influenza spread that integrates data gathered from different sources (Census, Surveys, Air Traffic). We plan to use an age-structured metapopulation model, with demographic and mobility data gathered form Census and social contact data obtained from surveys based on POLYMOD (FP5-6) approach [4]. The model will simulate the spread of influenza at the municipality scale, accounting for fluctuation of social contacts and mobility fluxes between weeks and week-ends and between school term and period of school closure. In order to build this country-level data-driven meta-population model, for both Belgium and France we will first gather and analyse socio-economic, contact and mobility data from different sources, namely, Census, mobility surveys, health surveillance, ministries of education, etc. These steps are country-specific and will result in having for each country the georeferenced population breakdown into children and adult classes, the school calendar, the mobility network and the contact matrices for regular week, regular week-ends, holiday weeks and holiday week-ends. We will then integrate the data to build the meta-population model. We will start with the case of Belgium, given the smaller country size, reduced complexity of its mobility network and the ready-to use social contact data analysed by the Belgian team. The code will then be adapted to the French case, integrating the French data and considering additional layers (such as e.g. the air transportation network).

Keywords:EPIDEMIOLOGY AND PUBLIC HEALTH, MODELS INFECTIOUS DISEASES, QUANTITATIVE RISK ASSESSMENT

Disciplines:Scientific computing, Bioinformatics and computational biology, Public health care, Public health sciences, Public health services