Prediction Model for Influenza Epidemic Based on Twitter Data

Abstract: Today controlling the influenza outbreak has become an important issue of health authorities worldwide to get rid of the epidemic as early as possible. In this research work we have done an associated study of algorithms and methods, modelling the outbreak of any epidemic with the focus of swine flu which must be prevented at an early stage if spread. In the Introduction section we have given the significance of the study with respect to micro-blogging websites like Twitter, Facebook, etc. that studies Social media platforms. In Related Work, we have done a survey from different resources and ideas applied to predict and detect the outbreak of epidemics and studied their advantages and limitations. After that new idea is proposed which can overcome the limitations of models have been proposed. This proposed model comprises a machine learning technique in order to make a model trained and we proposed a new idea of Swine Epidemic Hint algorithm which will look after epidemic activities happening on the Twitter and the Markov Chain state model to categorize epidemic activities into three stages (Beginning of Epidemic, Spread of Epidemic, Decay of Epidemic). Finally, we have proposed a new framework to model epidemic prediction based on the scope of improvements of previous work done.

Keywords: Twitter APIs, Markov Chain State Model, BOWs, Time series classification.