Data-centric models of COVID-19 have been attempted, but have certain limitations. In this work, we propose an agent-based model of the epidemic in a confined space of agents representing humans. An extension to the SEIR model allows us to consider the difference between the appearance (black-box view) of the spread of disease and the real situation (glass-box view). Our model allows for simulations of lockdowns, social distancing, personal hygiene, quarantine, and hospitalization, with further considerations of different parameters, such as the extent to which hygiene and social distancing are observed in a population. Our results provide qualitative indications of the effects of various policies and parameters, for instance, that lockdowns by themselves are extremely unlikely to bring an end to an epidemic and may indeed make things worse, that social distancing is more important than personal hygiene, and that the growth of infection is significantly reduced for moderately high levels of social distancing and hygiene, even in the absence of herd immunity.

中文翻译：

---

建模 COVID-19 的传播和控制

已经尝试了 COVID-19 的以数据为中心的模型，但有一定的局限性。在这项工作中，我们在代表人类的代理的密闭空间中提出了一种基于代理的流行病模型。SEIR 模型的扩展使我们能够考虑疾病传播的外观（黑盒视图）与真实情况（玻璃盒视图）之间的差异。我们的模型允许模拟封锁、社交距离、个人卫生、隔离和住院，并进一步考虑不同的参数，例如在人群中观察到的卫生和社交距离的程度。我们的结果提供了各种政策和参数影响的定性指示，例如，封锁本身极不可能结束流行病，并且确实可能​​使情况变得更糟，