Exploring the influence of human mobility factors and spread prediction on early COVID-19 in the USA

BACKGROUND: COVID-19 is still spreading rapidly around the world. In this context, how to accurately predict the turning point, duration and final scale of the epidemic in different countries, regions or cities is key to enabling decision makers and public health departments to formulate intervention measures and deploy resources. METHODS: Based on COVID-19 surveillance data and human mobility data, this study predicts the epidemic trends of national and state regional administrative units in the United States from July 27, 2020, to January 22, 2021, by constructing a SIRD model considering the factors of “lockdown” and “riot”. RESULTS: (1) The spread of the epidemic in the USA has the characteristics of geographical proximity. (2) During the lockdown period, there was a strong correlation between the number of COVID-19 infected cases and residents’ activities in recreational areas such as parks. (3) The turning point (the point of time in which active infected cases peak) of the early epidemic in the USA was predicted to occur in September. (4) Among the 10 states experiencing the most severe epidemic, New York, New Jersey, Massachusetts, Texas, Illinois, Pennsylvania and California are all predicted to meet the turning point in a concentrated period from July to September, while the turning point in Georgia is forecast to occur in December. No turning points in Florida and Arizona were foreseen for the forecast period, with the number of infected cases still set to be growing rapidly. CONCLUSIONS: The model was found accurately to predict the future trend of the epidemic and can be applied to other countries. It is worth noting that in the early stage there is no vaccine or approved pharmaceutical intervention for this disease, making the fight against the pandemic reliant on non-pharmaceutical interventions. Therefore, reducing mobility, focusing on personal protection and increasing social distance remain still the most effective measures to date.