Institute of Tropical Medicine, Eberhard Karls University, University Clinics Tübingen, Wilhelmstr. 27, 72074, Tübingen, Germany.
Department of Pathology, Institute of Pathology and Neuropathology, Eberhard Karls University, University Clinics Tübingen, Liebermeisterstr. 8, 72076, Tübingen, Germany.
Epidemiol Infect. 2020 Nov 13;148:e284. doi: 10.1017/S0950268820002757.
Google's 'Community Mobility Reports' (CMR) detail changes in activity and mobility occurring in response to COVID-19. They thus offer the unique opportunity to examine the relationship between mobility and disease incidence. The objective was to examine whether an association between COVID-19-confirmed case numbers and levels of mobility was apparent, and if so then to examine whether such data enhance disease modelling and prediction. CMR data for countries worldwide were cross-correlated with corresponding COVID-19-confirmed case numbers. Models were fitted to explain case numbers of each country's epidemic. Models using numerical date, contemporaneous and distributed lag CMR data were contrasted using Bayesian Information Criteria. Noticeable were negative correlations between CMR data and case incidence for prominent industrialised countries of Western Europe and the North Americas. Continent-wide examination found a negative correlation for all continents with the exception of South America. When modelling, CMR-expanded models proved superior to the model without CMR. The predictions made with the distributed lag model significantly outperformed all other models. The observed relationship between CMR data and case incidence, and its ability to enhance model quality and prediction suggests data related to community mobility could prove of use in future COVID-19 modelling.
谷歌的“社区流动性报告”（CMR）详细说明了 COVID-19 疫情下活动和流动性的变化。因此，它们提供了一个独特的机会来研究流动性与疾病发病率之间的关系。目的是检验 COVID-19 确诊病例数量与流动性水平之间是否存在关联，如果存在，那么检验这些数据是否能增强疾病建模和预测。对全球各国的 CMR 数据与相应的 COVID-19 确诊病例数量进行了交叉相关分析。为每个国家的疫情建立了模型来解释病例数量。使用贝叶斯信息准则对比了使用数值日期、同期和分布式滞后 CMR 数据的模型。引人注目的是，西欧和北美的主要工业化国家的 CMR 数据与病例发病率之间呈负相关。对所有大陆的全面检查发现，除了南美洲以外，其他大陆都呈负相关。在建模时，CMR 扩展模型优于没有 CMR 的模型。分布式滞后模型的预测明显优于所有其他模型。CMR 数据与病例发病率之间的观察到的关系及其增强模型质量和预测的能力表明，与社区流动性相关的数据可能在未来的 COVID-19 建模中证明是有用的。
