Brussels, 13 August 2007
Traditional surveillance systems monitor death rates, the utilisation of health services (e.g. emergency room admissions, drug prescriptions), abnormal patterns, changing laboratory characteristics and exposure to risks related to the environment, food or animals. Yet, certain public or animal health threats may go unnoticed but can be reported in the local press or by other less known sources. The challenge is how to find them.
Using pre-defined keywords and combinations, MediSys crawls through the web and sorts information into three primary categories: "Diseases", "Bioterrorism", and "Other Threats". Articles or "hits" are then classified in more precise categories such as "AIDS-HIV", "Respiratory infections", "Avian flu", "Legionella", "Anthrax" or "Nuclear safety". Statistics are stored on the filtered categories and "breaking news" is detected in a given category. Based on the level of new articles retrieved and the detected keywords, an automatic alert can then be sent by e-mail and SMS to decision-makers who are on permanent standby. For example, in the recent case of the outbreak of foot and mouth disease in the UK, MediSys detected a sudden jump in news reports and automatically sent email and SMS alerts to Public Health officials across Europe.
Working with the University of Helsinki, an automatic incident detection system has also been put in place. This analyses English reports and extracts structured data on the number of cases, the location and the date. This then feeds an automatic incident database used by the Member States and the European Commission. Citizens can also have free access to this automatic information scanning tool, which includes alert statistics, articles in various languages, and e-mail alerts.
Public version of MediSys: http://medusa.jrc.it.