Data-assimilation and state estimation for contact-based spreading processes using the ensemble kalman filter: Application to COVID-19

•A contact-based extended SEIQR model is presented to model epidemic spreading in complex networks.•The model is validated with data from the spreading of COVID-19 in Germany, Mexico and the US.•Accurate predictions over long time periods, more then 150 days are achieved.•An ensemble Kalman Filter is proposed on validated for state estimation and data-assimilation to account for stochastic uncertainties. The main aim of the present paper is threefold. First, it aims at presenting an extended contact-based model for the description of the spread of contagious diseases in complex networks with consideration of asymptomatic evolutions. Second, it presents a parametrization method of the considered model, including validation with data from the actual spread of COVID-19 in Germany, Mexico and the United States of America. Third, it aims at showcasing the fruitful combination of contact-based network spreading models with a modern state estimation and filtering technique to (i) enable real-time monitoring schemes, and (ii) efficiently deal with dimensionality and stochastic uncertainties. The network model is based on an interpretation of the states of the nodes as (statistical) probability densities samples, where nodes can represent individuals, groups or communities, cities or countries, enabling a wide field of application of the presented approach.