Abstract

detection of irrelevant articles, alternative English search strategies for 15 diseases were tested against existing alerts. Precision was measured using Positive Predictive Value (PPV), defined as the probability of detected article to have public health relevance, where relevance was determined by EI experts (human filter) who evaluated 13,679 English articles scanned within 24 hours. EU national public health experts supported sources and interface revision. Results: Alternative search strategies successfully reduced false positives without reducing sensitivity (PPV from 15% to 30%). Best refinement of search strategies requires using fewer keywords (diseases basic definitions), adjusting their weights with an extensive use of negative values and using epidemiological terms in the keywords combinations. New alerts have been set up for English and translation to other EU languages is ongoing with the support of public health experts. Expansion of the sources and restructuring of the interface are underway based on EU Member States feedback. Conclusion: Results indicate that optimization of search strategies and revision of sources with the support of national public health experts can help MedISys reaching its potential. Plans for 2010 include extension of the analysis to all diseases and languages and capacity building within the EU to encourage knowledge and use of the revised system. ECDC plans to create customized versions for EU Member States (country editions) and, on request, special versions for mass gathering surveillance.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call