Social media can predict the COVID-19 epidemic in China

Abstract Background：To investigate whether social media data can predict the prevalence of the COVID-19 epidemic and to compare the differences between the information on COVID-19 reflected in social media and the officially published information. Methods：The Severe acute respiratory syndrome coronavirus 2 related data from Microblog (Sina Weibo) from November 30 to December 31，2022 were collected using a Weibo crawler with a total of 598,285 entries. Text clustering, spatio-temporal analysis and sentiment analysis were used to explore hot topics of public interest and describe COVID-19 related information in China. Result：The COVID-19 outbreak in China in December 2022 can be divided into four phases, with peak symptoms occurring from December 13 to December 28 and peak symptom frequency on December 15. Shanghai was less concerned with COVID-19 information, while Beijing was the first province to experience a peak in fever symptoms. Guangdong, the first province to have a liberalized outbreak, had a later peak of fever, and Sichuan had the most discussion about fever symptoms. It is thought that the infection spread from the central and western provinces and some northern provinces to surrounding areas. In late December, national concern about respiratory symptoms decreased, instead, discussion of ear, nose, and throat and systemic symptoms increased. The general emotion of the population was positive. Conclusion：Social media can predict the COVID-19 epidemic in China. The peak of COVID-19 symptoms in China in December 2022 was on December 15, one week before the peak of the officially confirmed data.