In this paper we develop an SEIR-type model of COVID-19, with account for two particular aspects: non-exponential distribution of incubation and recovery periods, as well as age structure of the population. For the mean-field model, which does not distinguish between different age groups, we demonstrate that including a more realistic Gamma distribution of incubation and recovery periods may not have an effect on the total number of deaths and the overall size of an epidemic, but it has a major effect in terms of increasing the peak numbers of infected and critical care cases, as well as on changing the timescales of an epidemic, both in terms of time to reach the peak, and the overall duration of an outbreak. In order to obtain more accurate estimates of disease progression and investigate different strategies for introducing and lifting the lockdown, we have also considered an age-structured version of the model, which has allowed us to include more accurate data on age-specific rates of hospitalisation and COVID-19 related mortality. Applying this model to three comparable neighbouring regions in the UK has delivered some fascinating insights regarding the effect of lockdown in regions with different population structure. We have discovered that for a fixed lockdown duration, the timing of its start is very important in the sense that the second epidemic wave after lifting the lockdown can be significantly smaller or larger depending on the specific population structure. Also, the later the fixed-duration lockdown is introduced, the smaller is the resulting final number of deaths at the end of the outbreak. When the lockdown is introduced simultaneously for all regions, increasing lockdown duration postpones and slightly reduces the epidemic peak, though without noticeable differences in peak magnitude between different lockdown durations.

在本文中，我们开发了 COVID-19 的 SEIR 型模型，其中考虑了两个特定方面：潜伏期和恢复期的非指数分布，以及人口的年龄结构。对于不区分不同年龄组的平均场模型，我们证明包括更现实的潜伏期和恢复期的 Gamma 分布可能不会对死亡总数和流行病的总体规模产生影响，但它在增加受感染和重症监护病例的峰值数量以及改变流行病的时间尺度方面具有重大影响，无论是达到峰值的时间还是爆发的总持续时间。为了更准确地估计疾病进展并研究引入和解除锁定的不同策略，我们还考虑了模型的年龄结构版本，这使我们能够包含更准确的特定年龄住院率数据和 COVID-19 相关的死亡率。将此模型应用于英国三个可比较的邻近地区，就封锁对不同人口结构地区的影响提供了一些有趣的见解。我们发现，对于一个固定的封城时间，封城开始的时机非常重要，解封后的第二波疫情可能会小很多，也可能大很多，这取决于具体的人口结构。此外，越晚引入固定期限锁定，较小的是爆发结束时最终的死亡人数。当对所有地区同时实施封锁时，增加封锁持续时间会推迟并略微降低流行病高峰，但不同封锁持续时间之间的峰值幅度没有明显差异。