Original PaperDetecting Disease Outbreaks in Mass Gatherings Using InternetDataElad Yom-Tov 1 , BSc, MA, PhD; Diana Borsa 2 , BSc, BMath, MSc (Hons); Ingemar J Cox 3,4 , BSc, PhD; Rachel AMcKendry 5 , BSc, PhD1 Microsoft Research Israel, Herzelia, Israel2 Centre of Computational Statistics and Machine Learning (CSML), Department of Computer Science, University College London, University ofLondon, London, United Kingdom3 Copenhagen University, Department of Computer Science, Copenhagen, Denmark4 University College London, University of London, Department of Computer Science, London, United Kingdom5 University College London, University of London, London Centre for Nanotechnology and Division of Medicine, London, United KingdomCorresponding Author:Diana Borsa, BSc, BMath, MSc (Hons)Centre of Computational Statistics and Machine Learning (CSML)Department of Computer ScienceUniversity College London, University of LondonMalet PlaceGower StLondon, WC1E 6BTUnited KingdomPhone: 44 20 7679Fax: 44 20 7387 1397Email: d.borsa@cs.ucl.ac.ukAbstractBackground: Mass gatherings, such as music festivals and religious events, pose a health care challenge because of the risk oftransmission of communicable diseases. This is exacerbated by the fact that participants disperse soon after the gathering,potentially spreading disease within their communities. The dispersion of participants also poses a challenge for traditionalsurveillance methods. The ubiquitous use of the Internet may enable the detection of disease outbreaks through analysis of datagenerated by users during events and shortly thereafter.Objective: The intent of the study was to develop algorithms that can alert to possible outbreaks of communicable diseases fromInternet data, specifically Twitter and search engine queries.Methods: We extracted all Twitter postings and queries made to the Bing search engine by users who repeatedly mentioned oneof nine major music festivals held in the United Kingdom and one religious event (the Hajj in Mecca) during 2012, for a periodof 30 days and after each festival. We analyzed these data using three methods, two of which compared words associated withdisease symptoms before and after the time of the festival, and one that compared the frequency of these words with those ofother users in the United Kingdom in the days following the festivals.Results: The data comprised, on average, 7.5 million tweets made by 12,163 users, and 32,143 queries made by 1756 users fromeach festival. Our methods indicated the statistically significant appearance of a disease symptom in two of the nine festivals.For example, cough was detected at higher than expected levels following the Wakestock festival. Statistically significant agreement(chi-square test, P<.01) between methods and across data sources was found where a statistically significant symptom wasdetected. Anecdotal evidence suggests that symptoms detected are indeed indicative of a disease that some users attributed tobeing at the festival.Conclusions: Our work shows the feasibility of creating a public health surveillance system for mass gatherings based on Internetdata. The use of multiple data sources and analysis methods was found to be advantageous for rejecting false positives. Furtherstudies are required in order to validate our findings with data from public health authorities.(J Med Internet Res 2014;16(6):e154) doi:10.2196/jmir.3156J Med Internet Res 2014 | vol. 16 | iss. 6 | e154 | p.1 http://www.jmir.org/2014/6/e154/(page number not for citation purposes)Yom-Tov et al JOURNAL OF MEDICAL INTERNET RESEARCHXSL • FORenderX
