Detecting flu transmission by social sensor in China

The control and prevention of transmissible diseases is an important public health concern. This paper gives a strategy to detect the transmission of contagion based on social network. With the popularity of the Weibo, a hybrid of Twitter and Facebook in China, people often posts his/her feelings when he/she or his/her friend suffers from disease. Thus, Weibo plays a social sensor role to surveill people's health status. By exploring more than 35.3 million tweets from Sina Weibo, we first filter the data according to the ILI (Influenza like Illness) definition and compare the online social network data with the statistics summery of China CDC (Center of Disease Control and Prevention) using cosine similarity and the standard deviation metrics. Dynamic Bayesian Network is applied to detect the outbreak and transmission of influenza in a certain city considering the co-location and the social ties. The experiment results show our methods can give effective suggestions to control and prevent the influenza epidemic in China.