Piezomagnetic vibration energy harvester with an amplifier
Résumé
Epidemiological models of a mechanistic nature are a powerful tool to investigate potential courses of evolution of infectious disease outbreaks, being widely used by computational epidemiologists, to carry out qualitative and quantitative studies for many decades. However, due to the extremely uncertain nature of the information that feeds these models (epidemiological parameters, initial conditions, etc.), as well as the limited horizon of predictability intrinsic to epidemic outbreaks where human behavior influences the evolution of epidemic dynamics (e.g. COVID-19), the use of uncertainty quantification (UQ) techniques is mandatory in any quantitative context that exploits such models as predictive tools. In this context, the present work presents a new framework for UQ that combines optimization via cross-entropy method and approximate Bayesian computation to calibrate a mechanistic epidemic model based on ordinary differential equations and propagate the underlying parametric uncertainties through it [1,2]. The new methodology inherits the good properties of the two techniques that are combined, having a great capacity to gain information and relatively low computational cost. The effectiveness of the new methodology is tested with the aid of a compartmental model, which includes asymptomatic and dead hospitalizations, and real data related to the COVID-19 outbreak in the city of Rio de Janeiro, Brazil.
Origine : Fichiers produits par l'(les) auteur(s)