Medrek M, Pastuszak Z
Maria Curie-Sklodowska University, Faculty of Economics, Department of Information Systems and Logistics, Pl. Marii Curie-Sklodowskiej 5, 20-031 Lublin, Poland.
Expert Syst Appl. 2021 Mar 15;166:114109. doi: 10.1016/j.eswa.2020.114109. Epub 2020 Oct 15.
The COVID-19 virus outbreak has affected most of the world in 2020. This paper deals with artificial intelligence (AI) methods that can address the problem of predicting scale, dynamics and sensitivity of the outbreak to preventive actions undertaken with a view to combatting the epidemic. In our study, we developed a cellular automata (CA) model for simulating the COVID-19 disease spreading. The enhanced infectious disease dynamics (Susceptible, Exposed, Infectious, and Recovered) model was applied to estimate the epidemic trends in Poland, France, and Spain. We introduced new parameters into the simulation framework which reflect the statistically confirmed dependencies such as age-dependent death probability, a different definition of the contact rate and enhanced parameters reflecting population mobility. To estimate key epidemiological measures and to predict possible dynamics of the disease, we juxtaposed crucial CA framework parameters to the reported COVID-19 values, e.g. length of infection, mortality rates and the reproduction number. Moreover, we used real population density and age structures of the studied epidemic populations. The model presented allows for the examination of the effectiveness of preventive actions and their impact on the spreading rate and the duration of the disease. It also shows the influence of structure and behavior of the populations studied on key epidemic parameters, such as mortality and infection rates. Although our results are critically dependent on the assumptions underpinning our model and there is considerable uncertainty associated with the outbreaks at such an early epidemic stage, the obtained simulation results seem to be in general agreement with the observed behavior of the real COVID-19 disease, and our numerical framework can be effectively used to analyze the dynamics and efficacy of epidemic containment methods.
2020年，新型冠状病毒肺炎（COVID-19）病毒的爆发影响了世界上大部分地区。本文探讨了人工智能（AI）方法，这些方法能够解决预测疫情规模、动态变化以及疫情对所采取预防措施的敏感性等问题，以便抗击疫情。在我们的研究中，我们开发了一种细胞自动机（CA）模型来模拟COVID-19疾病的传播。采用增强型传染病动力学（易感、暴露、感染和康复）模型来估计波兰、法国和西班牙的疫情趋势。我们在模拟框架中引入了新的参数，这些参数反映了经统计确认的相关性，如年龄依赖性死亡概率、接触率的不同定义以及反映人口流动性的增强参数。为了估计关键的流行病学指标并预测疾病可能的动态变化，我们将细胞自动机框架的关键参数与报告的COVID-19数值并列比较，例如感染时长、死亡率和繁殖数。此外，我们使用了所研究疫情地区的实际人口密度和年龄结构。所提出的模型能够检验预防措施的有效性及其对疾病传播速度和持续时间的影响。它还显示了所研究人群的结构和行为对关键疫情参数（如死亡率和感染率）的影响。尽管我们的结果严重依赖于支撑我们模型的假设，并且在疫情早期阶段与疫情相关存在相当大的不确定性，但所获得的模拟结果似乎与实际COVID-19疾病的观察行为总体一致，并且我们的数值框架可有效地用于分析疫情防控方法的动态变化和效果。
