[Skip to Navigation]
Comment & Response
April 2017

Uncertainties in Big Data When Using Internet Surveillance Tools and Social Media for Determining Patterns in Disease Incidence

Author Affiliations
  • 1School of Engineering, University of Melbourne, Parkville, Victoria, Australia
JAMA Ophthalmol. 2017;135(4):402. doi:10.1001/jamaophthalmol.2017.0138

To the Editor Deiner et al1 recently reported that the early detection of the spatial spread of epidemics may be possible by using online tools such as Google Trends to analyze online queries on disease symptoms appearing in Google searches and social media postings. The hypothesis was that the incidence of conjunctivitis could be predicted by prior data from internet search statistics on symptoms.2,3 The use of pattern recognition algorithms and data analytics is projected to become a common approach to detect, recognize, and classify patterns of disease and human behaviour.2 Advantages include early warnings and the localization of biosecurity threats and epidemics to enable timely intervention by medical authorities. We encourage the authors to comment on the need for the correlation of internet information acquired by search engines and social media to validate hard data such as electronic medical records and hospital admissions and medication sales, and how this might be accomplished over the next decade.

Add or change institution