DI-UMONS : Dépôt institutionnel de l’université de Mons

Recherche transversale
Rechercher
(titres de publication, de périodique et noms de colloque inclus)
2016-10-04 - Colloque/Article dans les actes avec comité de lecture - Anglais - 8 page(s)

Ben Souissi Souhir, Abed Mourad, El Hiki Lahcen , Fortemps Philippe , Pirlot Marc , "Categorizing the suitability of an alternative for a subject. An application to antibiotics prescription recommendation" in IEEE Symposium on Computational Intelligence in Multicriteria Decision-Making, ATHENS, GREECE

  • Codes CREF : Sciences de l'ingénieur (DI2000), Intelligence artificielle (DI1180), Modèles mathématiques d'aide à la décision (DI1151), Génie hospitalier (DI2656)
  • Unités de recherche UMONS : Génie des Procédés chimiques et biochimiques (F505)
  • Instituts UMONS : Institut des Sciences et du Management des Risques (Risques)

Abstract(s) :

(Anglais) Nowadays there is a real need to operate and link existing knowledge expressed by experts, in domains in which highly reliable recommendation systems are needed. This is especially true in the medical domain where knowledge sources are heterogeneous, since they are separately formed in different contexts. A major difficulty is to relate these sources together in a way that respects the specic medical recommendation requirements. Using MCDM (Multi-Criteria Decision Making) models can help in this aim. The general problem we address is to assess the suitability of an alternative (or a solution) for a given subject in a specific context. For instance, which antibiotic (alternative) should be prescribed to a patient (subject) who suffers from bacterial infection, taking into account characteristics of the patient such as allergies, renal problems, etc. We use a MCDM sorting method (MR-Sort with Veto, a variant of ELECTRE TRI), to categorize the pairs alternative-solution (e.g. antibioticpatient) according to their degree of suitability. The contextual knowledge (e.g. side-effects of antibiotics, characteristics of patient), structured in several ontologies, is linked to the assessment model through a semantic model. The approach is applied to the recommendation of antibiotic prescription, in collaboration with the EpiCura Hospital Center.