Yarsky P
United States Nuclear Regulatory Commission, United States of America.
Math Comput Simul. 2021 Jul;185:687-695. doi: 10.1016/j.matcom.2021.01.022. Epub 2021 Feb 13.
A Susceptible-Exposed-Infected-Removed (SEIR) model was developed to forecast the spread of the novel coronavirus (SARS-CoV-2) in the United States and the implications of re-opening and hospital resource utilization. The model relies on the specification of various parameters that characterize the virus and the population being modeled. However, several of these parameters can be expected to vary significantly between states. Therefore, a genetic algorithm was developed that adjusts these population-dependent parameters to fit the SEIR model to data for any given state.
Publicly available data was collected from each state in terms of the number of positive COVID-19 cases and the number of COVID-19-caused deaths and used as inputs into a SEIR model to predict the spread of COVID infections in a given population. A genetic algorithm was designed where the genes are the state-dependent parameters from the model. The algorithm operates by determining the fitness of a given set of genes, applying selection, using selected agents to reproduce with cross-over, applying random mutation, and simulating several generations.
Use of the genetic algorithm produces exceptionally good agreement between the model and available data. Deviations in the parameters were examined to see if the trends were reasonable.
开发了一种易感-暴露-感染-清除（SEIR）模型，以预测新型冠状病毒（SARS-CoV-2）在美国的传播情况以及重新开放和医院资源利用的影响。该模型依赖于表征病毒和被建模人群的各种参数的设定。然而，可以预期其中几个参数在不同州之间会有显著差异。因此，开发了一种遗传算法，用于调整这些依赖于人群的参数，以使SEIR模型适用于任何给定州的数据。
从每个州收集公开可用的数据，包括新冠病毒检测呈阳性的病例数和新冠病毒导致的死亡人数，并将其用作SEIR模型的输入，以预测给定人群中新冠病毒感染的传播情况。设计了一种遗传算法，其中基因是模型中依赖于州的参数。该算法通过确定给定一组基因的适应度、应用选择、使用选定的个体进行交叉繁殖、应用随机突变以及模拟几代来运行。
遗传算法的使用使模型与现有数据之间达成了非常好的一致性。对参数偏差进行了检查，以确定趋势是否合理。
