Using Friends as Sensors to Detect Global-Scale Contagious Outbreaks


Online data is a rich source for examining epidemics and patterns across the globe. This study uses information from Twitter to examine local network structures and its impact on the contagious spread of information globally.

The Issue:

Due to the availability of data on a global scale, there exists an opportunity to study central individuals within human social networks and the ability to detect contagious outbreaks before they happen in the larger population.

Key Findings

  • The friend group was more central in the network than the control group.

  • The friend group helped detect viral outbreaks of novel hashtag use about seven days earlier than with an equally-sized randomly chosen group.

  • The study found that this method of contagious spread of information in a global-scale network works better than expected due to network structure--highly central actors are more active and more diverse in the information they share.


While the ability of the sensor method to detect outbreaks early depends on factors such as online context, the type of phenomenon spreading, and the size of the population, overall, local monitoring is a promising way to monitor contagious processes across global networks.

About the Study:

Six months of data from 2009 Twitter data were analyzed. A group of randomly chosen individuals from Twitter were sampled, along with a randomly chosen group of “friends” of members of this group. This created a control group and a sensor group with a higher degree of centrality.