ABSTRACT

This paper deals with the quantification of the different rates in epidemiological models from a function estimation framework, with the objective of identifying the desired unknowns without defining a priori basis functions for describing its behaviour. This approach is used to analyze data for the Covid-19 pandemic in Italy and Brazil. The forward problem is written in terms of the SIRD model, while the inverse problem is solved by combining the Levenberg–Marquardt method with Tikhonov regularization. A very good agreement was achieved between data and the calculated values. The resulting methodology is robust and very versatile, being easily applicable to other epidemiology models and data from other countries.

摘要

本文从函数估计框架处理流行病学模型中不同比率的量化，目的是识别所需的未知数，而无需定义先验基函数来描述其行为。这种方法用于分析意大利和巴西 Covid-19 大流行的数据。前向问题是根据 SIRD 模型编写的，而逆问题是通过结合 Levenberg-Marquardt 方法和 Tikhonov 正则化来解决的。数据和计算值之间达到了很好的一致性。由此产生的方法是稳健且非常通用的，很容易适用于来自其他国家的其他流行病学模型和数据。