Title: Estimating household contact matrices structure from easily collectable metadata

Abstract: The ability of developing accurate models plays a pivotal role in understanding infectious diseases dynamics and designing effective measures for epidemic mitigation. Contact matrices are of crucial importance to quantify the interaction between age groups, but their estimation is a rather time and effort consuming task. In this article we show that they can reliably be estimated combining a context-independent model with easily collectable data, such as co-presence and an age activity parameter. The proposed model is tested on high resolution proximity data collected in a rural and an urban village in South Africa. Given its simplicity and interpretability, we expect our method to be easily applied to other contexts as well and we identify relevant questions that need to be addressed during the data collection procedure.