In the paper, we propose a semiparametric framework for modeling the COVID-19 pandemic. The stochastic part of the framework is based on Bayesian inference. The model is informed by the actual COVID-19 data and the current epidemiological findings about the disease.

The framework combines many available data sources (number of positive cases, number of patients in hospitals and in intensive care, etc.) to make outputs as accurate as possible and incorporates the times of non-pharmaceutical governmental interventions which were adopted worldwide to slow-down the pandemic. The model estimates the reproduction number of SARS-CoV-2, the number of infected individuals and the number of patients in different disease progression states in time. It can be used for estimating current infection fatality rate, proportion of individuals not detected and short term forecasting of important indicators for monitoring the state of the healthcare system. With the prediction of the number of patients in hospitals and intensive care units, policy makers could make data driven decisions to potentially avoid overloading the capacities of the healthcare system. The model is applied to Slovene COVID-19 data showing the effectiveness of the adopted interventions for controlling the epidemic by reducing the reproduction number of SARS-CoV-2. It is estimated that the proportion of infected people in Slovenia was among the lowest in Europe (0.350%, 90% CI [0.245–0.573]%), that infection fatality rate in Slovenia until the end of first wave was 1.56% (90% CI [0.94–2.21]%) and the proportion of unidentified cases was 88% (90% CI [83–93]%).

The proposed framework can be extended to more countries/regions, thus allowing for comparison between them. One such modification is exhibited on data for Slovene hospitals.

在本文中，我们提出了一个用于对 COVID-19 大流行进行建模的半参数框架。该框架的随机部分基于贝叶斯推理。该模型基于实际的 COVID-19 数据和有关该疾病的当前流行病学发现。

该框架结合了许多可用的数据源（阳性病例数量、医院和重症监护室的患者数量等），使输出尽可能准确，并纳入了世界范围内为减缓疫情而采取的非药物政府干预措施的时间。压倒疫情。该模型及时估计了SARS-CoV-2的繁殖数量、感染个体的数量以及处于不同疾病进展状态的患者数量。它可用于估计当前感染死亡率、未检测到的个体比例以及监测医疗保健系统状态的重要指标的短期预测。通过预测医院和重症监护病房的患者数量，政策制定者可以做出数据驱动的决策，以避免医疗保健系统的容量超载。该模型应用于斯洛文尼亚 COVID-19 数据，显示了所采取的干预措施通过减少 SARS-CoV-2 繁殖数量来控制疫情的有效性。据估计，斯洛文尼亚的感染者比例是欧洲最低的（0.350%，90% CI [0.245–0.573]%），截至第一波浪潮结束时斯洛文尼亚的感染死亡率为1.56%（90%） CI [0.94–2.21]%），不明病例比例为 88%（90% CI [83–93]%）。

拟议的框架可以扩展到更多国家/地区，从而可以在它们之间进行比较。斯洛文尼亚医院的数据显示了这样的修改之一。