15.2.2 Project argus (georgetown university)
Since 2004, Project Argus, based at Georgetown University Medical Center,
has been working in the detection and tracking of biological events outside of
the United States. Project Argus is a mixed-mode system with an automatic
analysis component and a large team of approximately 40 human analysts.
The U.S. Centers for Disease Control and Prevention (CDC) and other fed-
eral agencies oversee health surveillance within the United States (Wilson
2007). A novel feature of Project Argus has been its pioneering use of early
warning indicators. Approximately 200 have been identified and incorpo-
rated into a staging model analogous to the U.S. National Weather Service's.
Categories exist for unifocal events, multifocal events, severe infrastructure
strain, social collapse, and recovery.
Project Argus covers approximately 40 languages, 29 of which are auto-
matically indexed by the article retrieval system, and 13 of which are machine
translated. The languages were chosen based on the availability of source
texts and regional analysts. Coverage of diseases is focused on approximately
150 fast-spreading diseases that affect humans, animals, and plants.
The system's automatic component ingests news directly from the Web
using a set of constantly maintained wrappers. Documents are initially
selected based on weighted keywords within an indicator and warning
ontology. A score is generated that reflects its public health severity. Human
analysts have access to both the automatic alerts as well as output from their
own standing searches on the news database. Working in regional teams, the
automated component identifies significant stories that are synthesized into
daily reports for the population of users.
15.2.3 MediSys (Joint research Centre, italy)
The European Commission's Medical Information System (MedISys) is a fully
automatic GHIS developed and run by the Joint Research Centre (JRC) in Ipsa,
Italy (Steinberger et al. 2008). Working in close collaboration with the Health
Threats Unit at the Directorate General for Health and Consumer Protection
in the European Commission (DG SANCO), MedISys offers early warning of
i n fe c t iou s d i s e a s e, c he m ic a l, rad iolog ic a l, a nd nuc le a r he a lt h t h r e at s to EU i n st i-
tutions and member state organizations. Since 2007, MedISys has been working
in collaboration with the University of Helsinki's Pattern-Based Understanding
and Learning System (PULS), which we describe in the next section. PULS
performs precise linguistic analysis on document contents. Results from both
systems appear together on a common portal site run by MedISys.
In total, approximately 40 languages are covered by MedISys, including
all European Union languages. News reports in 26 of these languages are
made available on the MedISys Web portal. An extensive range of public
health threats is covered by the system and categorized into approximately
300 health-related categories in a multilingual ontology. Geographically,