CONICET | Buscador de Institutos y Recursos Humanos

To represent the complex individual interactions in the dynamics of disease spread informedby data, the coupling of an epidemiological agent-based model with the ensemble Kalmanfilter is proposed. The statistical inference of the propagation of a disease by means ofensemble-based data assimilation systems has been studied in previous works. The modelsused are mostly compartmental models representing the mean field evolution through ordinarydifferential equations. These techniques allow to monitor the propagation of the infectionsfrom data and to estimate several parameters of epidemiological interest. However,there are many important features which are based on the individual interactions that cannotbe represented in the mean field equations, such as social network and bubbles, contacttracing, isolating individuals in risk, and social network-based distancing strategies. Agentbasedmodels can describe contact networks at an individual level, including demographicattributes such as age, neighborhood, household, workplaces, schools, entertainmentplaces, among others. Nevertheless, these models have several unknown parameterswhich are thus difficult to prescribe. In this work, we propose the use of ensemble-baseddata assimilation techniques to calibrate an agent-based model using daily epidemiologicaldata. This raises the challenge of having to adapt the agent populations to incorporate theinformation provided by the coarse-grained data. To do this, two stochastic strategies to correctthe model predictions are developed. The ensemble Kalman filter with perturbed observationsis used for the joint estimation of the state and some key epidemiologicalparameters. We conduct experiments with an agent based-model designed for COVID-19and assess the proposed methodology on synthetic data and on COVID-19 daily reportsfrom Ciudad Auto´noma de Buenos Aires, Argentina.

enviar mensaje