Department of Data-centric Problem Solving Research, Korea Institute of Science and Technology Information, Daejeon, Republic of Korea.
Center for Convergent Research of Emerging Virus Infection, Korea Research Institute of Chemical Technology, Daejeon, Republic of Korea.
PLoS One. 2020 Nov 4;15(11):e0241466. doi: 10.1371/journal.pone.0241466. eCollection 2020.
As the number of global coronavirus disease (COVID-19) cases increases, the number of imported cases is gradually rising. Furthermore, there is no reduction in domestic outbreaks. To assess the risks from imported COVID-19 cases in South Korea, we suggest using the daily risk score. Confirmed COVID-19 cases reported by John Hopkins University Center, roaming data collected from Korea Telecom, and the Oxford COVID-19 Government Response Tracker index were included in calculating the risk score. The risk score was highly correlated with imported COVID-19 cases after 12 days. To forecast daily imported COVID-19 cases after 12 days in South Korea, we developed prediction models using simple linear regression and autoregressive integrated moving average, including exogenous variables (ARIMAX). In the validation set, the root mean squared error of the linear regression model using the risk score was 6.2, which was lower than that of the autoregressive integrated moving average (ARIMA; 22.3) without the risk score as a reference. Correlation coefficient of ARIMAX using the risk score (0.925) was higher than that of ARIMA (0.899). A possible reason for this time lag of 12 days between imported cases and the risk score could be the delay that occurs before the effect of government policies such as closure of airports or lockdown of cities. Roaming data could help warn roaming users regarding their COVID-19 risk status and inform the national health agency of possible high-risk areas for domestic outbreaks.
随着全球冠状病毒病 (COVID-19) 病例数量的增加，输入病例的数量逐渐上升。此外，国内疫情也没有减少。为了评估韩国输入性 COVID-19 病例的风险，我们建议使用每日风险评分。使用约翰霍普金斯大学中心报告的确诊 COVID-19 病例、韩国电信收集的漫游数据以及牛津 COVID-19 政府反应追踪器指数来计算风险评分。风险评分与 12 天后的输入性 COVID-19 病例高度相关。为了预测韩国 12 天后的每日输入性 COVID-19 病例，我们使用简单线性回归和自回归综合移动平均（ARIMA）开发了包含外生变量（ARIMAX）的预测模型。在验证集中，使用风险评分的线性回归模型的均方根误差为 6.2，低于不使用风险评分作为参考的 ARIMA（22.3）。使用风险评分的 ARIMAX 的相关系数（0.925）高于 ARIMA（0.899）。输入病例和风险评分之间出现 12 天时间滞后的一个可能原因是，机场关闭或城市封锁等政府政策的效果可能存在延迟。漫游数据可以帮助漫游用户了解其 COVID-19 风险状况，并向国家卫生机构通报国内疫情可能出现的高风险地区。
