This paper proposes a data-driven machine learning framework for parameter estimation and uncertainty quantification in epidemic models based on two key ingredients: (i) prior parameters learning via the cross-entropy method and (ii) update of the model calibration and uncertainty propagation through approximate Bayesian computation. The effectiveness of the new methodology is illustrated with the aid of actual data from COVID-19 epidemic at Rio de Janeiro city in Brazil, employing an ordinary differential equation-based model with a generalized SEIR-type mechanistic structure that includes time-dependent transmission rate, asymptomatics, and hospitalizations. A minimization problem with two cost terms (number of hospitalizations and deaths) is formulated, and twelve parameters are identified. The calibrated model provides a consistent description of the available data, able to extrapolate forecasts over a few weeks, which makes the proposed methodology very appealing for use in the context of real-time epidemic modeling.
翻译:本文提出基于两个关键要素的流行病模型参数估计和不确定性量化数据驱动机学习框架:(一) 以往的参数学习,通过交叉热带方法进行学习,(二) 通过近似贝叶斯计算更新模型校准和不确定性传播情况,巴西里约热内卢市COVID-19流行病的实际数据有助于说明新方法的有效性,采用普通的差别方程模型,采用通用的SEIR型机械结构,包括基于时间的传播率、无症状和住院。制定了两个成本条件(住院和死亡人数)的最小化问题,并确定了12项参数。校准模型对可用数据进行了一致描述,能够对几周的预测进行外推,从而使拟议方法非常需要用于实时流行病模型。