Architecture et standardisation des systèmes de surveillance épidémiologique : application au réseau sentinelles

The objective of this thesis was to show up how public health surveillance integrates evolutions of information technologies. A literature review first presents the state of the art of data sources used by public health surveillance systems for human disease surveillance, the architecture used to build such system the treatments applied to data in order to assess the surveillance goal and finally the data standardization in the field to exchange input and output data of these systems. A second part presents our work in the field:1 – JSentinel, is a desktop-based generic data collection tool based on an abstract data model and communication protocol. An evaluation of its usage by GPs of the French Sentinelles network showed a reduction of the notification delay and an increase of the rate of provided individual data for each notified case. 2 – Two solutions were compared to standardize exchanges of data from PHS systems: SDMX-HD (Statistical Data and Metadata Exchange) and OData. These two solutions were implemented to publish French Sentinelles network’s data. SDMX-HD based solution, by promoting an harmonization of metadata and data structure leads to a better interoperability and the OData solution, supported by a wide ecosystem allows to easily publish data but does not propose a common data model. Sustainability of these standards is also a blocking factor to promote them as a standard candidate. 3 – In a third work we used query log data from the Google‘s search engine as surveillance data for several infectious diseases and compare it to the French Sentinelles network’s incidence rates. Finally the results and future evolutions are discussed.