Top content

Why flu-prediction?

Combination of fast but fuzzy Social media data, and slow and precise data from the CDC

Including social media data reduces the prediction delay by 8 days

Our Project with Watson: Flu-Prediction

Our project on flu prediction (www.flu-prediction.com) demonstrates the power of cognitive computing and its potential to change the world. Flu is an infectious disease with a high societal and economical relevance. Flu threat management requires reliable predictions of the flu threat and fully informed people who can react adequately. We present FluPrediction.com, a Watson/Bluemix-based cognitive system that uses CDC data on flu infections, fused with social media data, to have an anytime and real-time prediction, and to identify structure and speed of flu spreading in the United States. In combination with the Watson Engagement Advisor that supports the user with important flu-related information, this cognitive system enables the user to react proactively and optimally, given the individual situational context.

www.flu-prediction.com is a great example of cognitive computing using efficient tools, demonstrating the efficiency and power of Bluemix tools. It was developed by a group of three master’s students of cognitive science over a period of 5 months only. It integrates several methods from artificial intelligence, natural language processing and machine learning, into a single advanced cognitive assistant that engages with the user intuitively and via natural language. To this end, it uses natural language processing to understand tweets, and it searches the content of the up to 500 million twitter tweets a day, classifying these in real-time, based on Bluemix services, in terms of flu relevance. It uses advanced cognitive analytics to predict flu outbreak based on the twitter data and fuses this with CDC predictions. And most importantly, it engages the user in closed loop interaction to provide background information that enables the user to be optimally prepared and to be ready to act optimally, based on Watson Engagement Advisor. The result is a fully informed user who can understand the flu development, and therefore know how to protect and act. The user can react proactively and be a part of the protection strategy. This makes this cognitive system a new experience. A cognitive system as a human-like super expert that enables the user to act and understand.