Avian Influenza Risk Surveillance in North America with Online Media.

摘要: The use of Internet-based sources information for health surveillance applications has increased in recent years, as a greater share social and media activity happens through online channels. potential value about emergent events include early warning, situational awareness, risk perception evaluation messaging among others. challenge harnessing these data is the vast number to monitor developing tools translate dynamic unstructured content into actionable information. In this paper we investigated one outlet, Twitter, avian influenza North America. We collected AI-related messages over five-month period compared official records AI outbreaks. A fully automated extraction analysis pipeline was developed acquire, structure, analyze an context. Two methods outbreak detection; static threshold cumulative-sum threshold; based on time series model normal were evaluated their ability discern important periods activity. Our findings show that peaks related real-world events, with outbreaks Nigeria, France USA receiving most attention while those China less evident data. Topic models found themes specific method, many method ambiguous. Further analyses might focus quantifying bias coverage relation between characteristics detectability Finally, here focused broad trends, there likely additional identifying low-frequency messages, operationalizing methodology comprehensive system visualizing patterns extracted from Internet, integrating other such wildlife, environment, agricultural