The ability to anticipate an epidemic would be invaluable for public health authorities. A new model-- published in PLOS Computational Biology by researchers: Researchers Nicholas Generous, Geoffrey Fairchild, Alina Deshpande, Sara Y. Del Valle, and Reid Priedhorsky-- could change the way we approach epidemic diseases.
The authors have recently used public data available from Wikipedia searches to model disease spread in 4 countries. They were able to effectively model influenza in the US, Thailand, Japan and Poland and Dengue in Brazil and Thailand. People tend to search their symptoms or research their disease before going into a hospital if they have a suspicion of having the disease. These trending searches can lead to predictive measurements up to 28-days before official governmental data.
The model has great strengths in that it is very adaptable country to country, however there are several hurdles in modeling some diseases and some countries. The model cannot yet predict the spread of rare diseases such as Ebola, or diseases with a longer incubation times such as HIV. Another hurdle is language, while Japanese is fairly country specific, English is widely spoken. Yet the researchers were still able to predict the spread of influenza in the US—they hope to receive country specific data in the future to improve the model further. Improvements on the model could certainly enlighten the spread of more difficult diseases in the future.
This is an exciting opportunity. To the authors’ knowledge, other computational models of diseases are limited to local contexts and limited to diseases, but this new model is widely transferrable. Because Wikipedia is free and open to the public, epidemics can be modeled in countries where official data is unavailable. Flu alone kills 3,000 to 49,000 people each year in the United States. Such a model could give us a leg up in preparing against disease both seasonal and emerging, likely will be an asset for public health in the near future.
-- Will St. Amant