GODS^N: Global News Driven Disease Outbreak and Surveillance

Abstract

Disease surveillance has evolved dramatically in the last few years, becoming more real-time, comprehensive and technology driven. As the Internet grows and evolves as a powerful information medium, the ability to mine real-time news feeds for disease surveillance has become viable. We propose a system, GODS^N that monitors global news for disease outbreaks and surveillance. The system processes real-time news feeds using natural language processing to obtain disease information and the geographical reference to plot them on a geographic information system. GODS^N provides an effective approach to visualize the spatial and temporal trends of infectious disease outbreaks or disease specific developments.

Background

Disease surveillance is a vital public health enterprise that is evolving tremendously as witnessed by the development of syndromic surveillance systems¹ that have narrowed the gap between event occurrence and detection. However, the surveillance systems are typically maintained by public health departments who often do not utilize the large amount of relevant information present on the web. We have developed GODS^N (read as n-GODs) that processes real-time news feeds to extract specific diseases (or healthcare) events and their location of occurrence. We use publicly available web services such as Google^TM Map² (geographic information system) and news aggregators² that use RSS (Really Simple Syndication) to collect real-time news feeds from thousands of news sources around the world. The system is currently an early-prototype and under active development.

System Architecture

Using open application programming interfaces (APIs), news feeds are obtained from a news aggregator² and then passed through a filter to extract health-related stories (Figure 1). A natural language processing system (MetaMap³) processes the feeds to extract relevant concepts such as disease names and the reference to a geographic location. The Web interface of the system (Figure 2) allows users to query for diseases and view their temporal and spatial evolution (Figure 3).

Figure 1.

System Architecture of the GODS^N system

Figure 2.

The search results for ‘bird flu’ on news feeds

Figure 3.

Temporal snapshots of the recent spread of ‘bird flu’

Conclusion

Dynamic information available on the web has a huge potential to expand the scope of disease surveillance. The future goal for GODS^N system is to create a repository of geographically tagged health news, stories and consumer driven health content.