Towards early detection of influenza epidemics by using social media analytics

Kwang Deok Kim, Liaquat Hossain

Research output: Chapter in Book/Report/Conference proceedingConference Paper published in Proceedingspeer-review

Abstract

This study aims at the possibility of early detection of seasonal influenza causing a major public health concern each year worldwide. Furthermore, the viability of social media analytics for the early detection of disease outbreak is examined. To conduct an analysis, three sets of data are used and investigated; search query data and two influenza related data collected from the nine US census bureau divisions on a weekly basis during a period from week 1, 2012 to week 52, 2013 from Google and Centers for Disease Control and Prevention (CDC), respectively. Pearson's correlation coefficient is calculated among the data as a statistical measure of the strength of a relationship, and the significant Pearson's rho confirms the existence of a very strong positive correlation. This study argues that the relative frequency of search queries is highly correlated with the actual number of influenza activity in each division of the US, and that social media analytics may be utilized to make an early detection of influenza epidemics possible. The importance of a detailed analysis is also discussed to assess the statistical methods for the phenomenon because the data from Google is still in its early stage which may contain inaccuracies.

Original languageEnglish
Title of host publicationDSS 2.0 - Supporting Decision Making with New Technologies
Place of PublicationNetherlands, United States
PublisherIOS Press
Pages36-41
Number of pages6
ISBN (Print)9781614993988
DOIs
Publication statusPublished - 1 Jan 2014
Externally publishedYes

Publication series

NameFrontiers in Artificial Intelligence and Applications
Volume261
ISSN (Print)0922-6389

Fingerprint Dive into the research topics of 'Towards early detection of influenza epidemics by using social media analytics'. Together they form a unique fingerprint.

Cite this