This paper presents two applications for the breast cancer treatment decision helping. The first one is called Casimir/RBR and can be likened to a rule-based reasoning system. In some situations, the application of the rules of this system does not provide a satisfying treatment. Then, the application Casimir/CBR-which is not fully implementedcan be used. Casimir/CBR uses principles of case-based reasoning in order to suggest solutions by adapting the rules of Casimir/RBR. In this framework, the rules are considered as cases: they are adapted rather than used literally.

Oncolor, the French regional cancer network of Lorraine, proposes clinical guidelines for the diagnosis, the treatment and the follow-up of cancers. In addition to paper documents, two computerized supports, a web site and the Kasimir software have been developed to make the use of these guidelines and their updating easier. This study aims at evaluating these three tools in breast and prostate cancer and to determine if computerized supports provide an additional help and an added value. From May to July 2001, 25 physicians of the Oncolor network have analysed, by consulting the guidelines on each of the 3 supports, 23 clinical cases of breast cancer and 22 cases of prostate cancer, i.e. in total 1,293 cases. We observe for breast cancer a response rate consistent with the guideline significantly higher with the Kasimir software compared to paper (87.1% versus 77.8%) or to Internet (87.1% versus 75.8%), an optimisation of the recourse to a specialised physician or a pluridisciplinary committee and globally a good satisfaction of the users. These results induce us to continue with the Kasimir experiment, to develop it for other cancers, especially those with a complex management and to evaluate its impact on medical practices.

This paper describes the Kasimir project which aim is the knowledge management for the cancer treatment and which gathers researchers in computer science and in ergonomics, experts in cancerology and an association federating physicians from Lorraine involved in cancerology. For each cancer localisation, the treatment is based on a protocol. The system Kasimir-RBR implements this protocol. The system Kasimir-CBR will implement adaptations of the protocol for ``non-standard cases''. The third part of the project is to study how protocol adaptations can lead to evolutions of the protocol.

This paper describes the Kasimir project which aim is the knowledge management for the cancer treatment and which gathers researchers in computer science and in ergonomics, experts in cancerology and an association federating physicians from Lorraine involved in cancerology. For each cancer localisation, the treatment is based on a protocol. The system Kasimir-RBR implements this protocol. The system Kasimir-CBR will implement adaptations of the protocol for ``non-standard cases''. The third part of the project is to study how protocol adaptations can lead to evolutions of the protocol.

La phase d'adaptation du raisonnement a partir de cas s'appuie sur des connaissances qui ne sont pas toujours simples a mettre en evidence. Cet article presente et analyse une experience d'acquisition et de modelisation de connaissances d'adaptation dans le domaine de l'aide au traitement du cancer du sein. Il montre comment les notions relevant des chemins de similarite apparaissent comme des outils d'aide a ces acquisitions et modelisations.

Cet article presente le systeme Casimir, un systeme a base de connaissances en cours de conception et de developpement qui a pour objectif l'aide a la decision therapeutique dans le cadre du traitement du cancer du sein. Les connaissances sont representees par une hierarchie de classes definies par les proprietes utiles des patients (et de leurs tumeurs). A certaines de ces classes sont associes des traitements medicaux. Le raisonnement dit protocolaire consiste a classer le patient a traiter et les proprietes de son cancer dans cette hierarchie puis a en deduire les traitements a effectuer. Malheureusement, un tel raisonnement ne donne pas necessairement satisfaction. En effet, d'une part les classes dont on dispose sont insuffisantes pour traiter tous les cas et, d'autre part, meme si le raisonnement protocolaire fournit un traitement, ce traitement peut s'averer inapplicable. Par consequent, un raisonnement non protocolaire s'avere necessaire. Ce raisonnement consiste a adapter les classes considerees et les traitements associes au patient. Les connaissances mises en jeu pour l'adaptation sont destinees a etre integrees au protocole, avec l'accord des specialistes.