Digital Epidemiology: Use of Digital Data Collected for Non-epidemiological Purposes in Epidemiological Studies

Park, Hyeoun-Ae; Jung, Hyesil; On, Jeongah; Park, Seul Ki; Kang, Hannah

However, search queries of GT frequently overestimate the incidence of illness. According to research carried out by a team at Northeastern University and Harvard University, GFT forecasted twice as many influenza cases as actually occurred in the United States during the 2012-2013 flu season [5] . Furthermore, the estimates cannot be reproduced easily because Google data is not publicly available. Twitter became an alternative data source because anyone with an internet connection can retrieve Twitter data. A group of researchers used data from Twitter to track level of disease activity and concern about the influenza H1N1 pandemic in 2011 [6] . Twitter has also been used to assess health sentiments such as those about vaccination [7] and to monitor drug safety [8] . Wikipedia, another publicly accessible data source, has also been used for digital epidemiology. Researchers at Boston Children's Hospital introduced a method of estimating the level of ILIs, in near-real-time, in the United States by monitoring the rate of influenza-related Wikipedia article views on a daily basis [9] .