Background: The novel coronavirus disease (COVID-19) spread world widely and caused a new pandemic. The Chinese government took strong intervention measures in the early stage of the epidemic, including strict travel-ban and social distancing policies. Prioritizing contribution of different factors is important for precise prevention and control of infectious diseases. Here, we proposed a novel framework for resolving this question and applied it to data from China.
Objective: To systematically reveal factors and their contribution to the control of COVID-19 in China, both at the national and city level.
Methods: Daily COVID-19 cases and related multidimensional data, including travel-related, medical, socioeconomic, environmental, and Influenza-like illness factors, from 343 cities in China were collected. Correlation analysis and interpretable machine learning algorithm were used to explore the quantitative contribution of different factors on either new cases or growth rate of COVID-19 for the epidemic period from January 17 to February 29, 2020.
Results: Many factors considered in this study are correlated to the spread of COVID-19 in China. Overall, travel-related population movements are the main contributing factors for both new cases and growth rate of COVID-19 in China, and the contributions are as high as 77% and 41%, respectively. There is a clear lag effect for travel related factors (previous vs current week: 45% vs 32% for new cases, and 21% vs 20% for growth rate). The contribution for travel from non-Wuhan regions is non-ignorable (12% and 26% for new cases and growth rate), especially for the growth rate (rank first as a single factor). City flow, a measure of control strength, contributes 16% and 7% to new cases and growth rate. Socioeconomic factors also play important roles in the growth rate of COVID-19 in China with 28% contribution. Other factors, including medical, environmental and influenza-like illness ones, also contribute to new cases and growth rate of COVID-19 in China. Based on the analysis for individual city, compared to Beijing, population flow from Wuhan and internal flow within the city are driving factors for more new cases in Wenzhou, while for Chongqing the contribution is mainly from population flow from Hubei beyond Wuhan. The higher growth rate for Wenzhou is driven by its population-related factors.
Conclusions: Many factors contributed to the outbreak outcome of COVID-19 in China. Travel-related population movement was the main driving factor with strong lag effect, and population movement from non-Wuhan regions is a non-ignorable hidden variable. For the growth rate, more factors were involved, including the socioeconomic ones that contributed more than a quarter. Those differential effects for various factors, along with city-level specificity, emphasize the importance of targeted and precise strategies for outbreak control of current COVID-19 crisis and other future infectious diseases.