Les résultats du projet ARIANE nous ont incités et encouragés à poursuivre des développements en collaboration avec des partenaires industriels. Pour cela nous avons établi avec la fondation "Health On the Net", dirigée par le Pr J-R. Scherrer, une collaboration visant à potentialiser nos efforts et nos résultats respectifs. De cette concertation est né le projet WRAPIN.
WRAPIN était un projet européen ( IST-2001-33260 ) de 2001 à 2003.
HON Health On the Net Foundation
HUG - DIM Division of Medical Informatics - University Hospitals of Geneva
LERTIM Laboratoire d'Enseignement et de Recherche sur le Traitement de l'information Médicale
MISS Mission pour l'informatisation du Système de Santé
THALES Information Systems (formerly of Thomson)
XR Partner Information Systems
Collaboration with other organizations:
INPES Institut National de Prévention et d'Education pour la Santé
OESO Foundation Health Operating Polydisciplinary Expertise Services
It appears more and more obvious that state governments will not succeed in the mastering of Web information quality and that international institutions are even more unable than states to do so. To consider that it might become possible to set-up a world Web organisation that could play the role of peer-review evaluation of the best international journals looks completely unrealistic. Only automatic ways of evaluation could be taken into account. We have discussed that option with quite a few high-level professional librarians (dealing with healthcare libraries) and they fully share this concept. Indeed there is also a need for further HONcode enhancements. But the strategy that is sketched here will start from Medline MeSH terms to be automatically generated out of the Web information by modalities of natural language processing like the approach by i.e. “morpho-sémantèmes”. In order to forge functions of association of corresponding MeSH terms as searching keys to Medline, the procedure is supposed to go up to the level of identification of appropriate medical/health journals, with their corresponding impact factors and matching parts of journal summaries with the submitted web text. A similar approach still using the MeSH terms for building up search keys could be used also to automatically access the information of clinical trial registries (particularly those of the NLM and of the UK BioMedCentral as well as the UK BioMed registries). A tentative score of quality issued from these different actions should be calculated and the information then tentatively classified by quality levels. This scoring approach will have to be matched with evidence based medicine (EBM) existing rules at the HON level platform with the help of ARIANE.
Using extensively the natural language processing tools (NLP) there will be systematic evaluations of submitted medical/healthcare information to Medline and other similar DBs as well as the systematic use of national registries including those of the US dealing with the FDA trials.
In order to get more granularity of information to be suited to complex queries. It appears progressively that detailed professional databases have progressively to be also considered. Professional associations working with us will contribute specifically to this part.
Up to now the HONcode has been used by our Foundation together with a large selection of institutions – at present more than three thousand. In addition the eight principles of the HONcode itself help to unify and standardise information on the WWW. Indeed the problem is therefore no longer to find information but to assess the credibility of the data source as well as the relevance and correctness of a document retrieved from the Net. In many cases no appropriate documentation is provided regarding the scientific design of the medical study nor are studies made available that support given claims. So the first publication of the HONcode of conduct was made available in July 1996 (http://WWW.hon.ch/honcode/conductVs1_5.html). It is also true that webmasters dealing with us showed that they were seeking guidelines for content presentation and were willing to improve their sites by following some basic rules. The HONcode was therefore developed as a self-governing initiative that webmasters can accept or reject. By displaying the HONcode logo on a website, the owners of the site are demonstrating their desire to provide medical health information, services or resources via the WWW that are of genuine value to the Net community and that meet specific quality standards. At the present time there is already a MedHunt database where each subscribing site is registered and followed up. With the tremendous increase in customers and users new tools and functionalities will have to be added to MedHunt and to HONcode databases as well. Currently the HONcode policy consists of the registration into the MedHunt database along with a description of the Websites supporting this initiative and the use of tracking techniques applied to those sites. A future method of policy of this initiative is to provide to each website desiring to abide and follow the HONcode a unique random number association with validation of the questionnaire. A HONcode logo that is displayed without having been registered is launching a warning message. This method provides an automatic list of websites abiding to the HONcode along with the Websites URL and email addresses.
More information at: http://www.wrapin.org/