Towards the Semantic Interpretation of Personal Health Messages from Social Media

Recent attempts have been made to utilise social media platforms, such as Twitter, to provide early warning and monitoring of health threats in populations (i.e. Internet bio-surveillance). It has been shown in the literature that a system based on keyword matching that exploits social media messages could report flu surveillance well ahead of the Centers of Disease Control and Prevention (CDC). However, we argue that a simple keyword matching may not capture semantic interpretation of social media messages that would enable healthcare experts or machines to extract and leverage medical knowledge from social media messages. In this paper, we motivate and describe a new task that aims to tackle this technology gap by extracting semantic interpretation of medical terms mentioned in social media messages, which are typically written in layman's language. Achieving such a task would enable an automatic integration between the data about direct patient experiences extracted from social media and existing knowledge from clinical databases, which leads to advances in the use of community health experiences in healthcare services.