2011 IEEE International Conference on Fuzzy Systems June 27-30, 2011, Taipei, Taiwan 978-1-4244-7317-5/11/$26.00 ©2011 IEEE Case Based Fuzzy Cognitive Maps (CBFCM) : New method for medical reasoning Comparison study between CBFCM/FCM Nassim Douali Inserm UMRS 872 équipe 20 Cordeliers Research Center Paris, France Nassim.douali@crc.jussieu.fr Elpiniki I. Papageorgiou Dept. of Informatics & Computer Technology Technological Educational Institute of Lamia, Lamia Greece epapageorgiou@teilam.gr Jos De Roo Agfa HealthCare Agfa HealthCare NV, Moutstraat 100, 9000 Gent, Belgium Marie-Christine Jaulent Inserm UMRS 872 équipe 20 Cordeliers Research Center Paris, France Marie-christine.jaulent@crc.jussieu.fr Abstract— Doctor usually uses his experience from the clinical practice to confirm a diagnosis and to prescribe an appropriate treatment for a specific patient. The computerized medical reasoning should not only focus on existing medical knowledge but also on physician’s previous experiences and new knowledge. Such knowledge and experience are vague and define uncertain relationships between facts and diagnosis. Case Based Fuzzy Cognitive Maps (CBFCM) are proposed as an evolution of Fuzzy Cognitive Maps (FCM) that allow more complete representation of knowledge since case-based fuzzy rules are introduced to improve FCM decision support systems. Semantic web is used to implement both FCM approaches. A database of 71 patients with urinary tract infections was used to perform the proposed approach. A comparative study between FCM (92%) and CBFCM (99%) was conducted and the results derived by CBFCM approach showed CBFCM to be superior to FCM. Keywords : Fuzzy Cognitive Maps, Rules, Decision Support Systems, Knowledge Representation, Clinical Guidelines, Semantic Web, case based reasoning. I. INTRODUCTION In medical practice, the doctor evokes several diagnoses but the differential diagnosis remains the major problem in medical decision making. It requires a several complementary examinations and experts review. Clinical Decision Support Systems (CDSS) are tools to improve patient safety and care processes [1,2]. Several studies and reviews have shown their effectiveness, i.e. [3,4,5,6] in the areas of diagnosis and therapy. In a recent review, Peleg and others [7] claim that CDSS allow “right knowledge to the right people in the right form at the right time” [8]. The development of CDSS has two knowledge-management tasks: the possibility of integration into the care system workflow, and knowledge management for proper decision making. CDSS models have to take into consideration issues related to patient data quality that is: incompleteness, poorly structured and sometimes unreliable data, problems associated with inaccuracies and uncertainties in Clinical Practice Guidelines (CPG). Many clinical diagnosis tasks involve reasoning under uncertainty [9]. Modeling a dynamic system can be hard in a computational sense. The use of CDSS in urinary tract infections will reduce false diagnoses and establish a proper antibiotic regimen. FCMs have proved their efficiency in medical decision support and medical reasoning through the literature [10-14]. This paper describes and evaluates a new method to create a diagnosis model framework based on semantic web approach and Case Based Fuzzy Cognitive Maps (CBFCM). The produced results of the examined framework are compared with those obtained from FCM reasoning approach for 71 patients. II. BACKGROUND A. Clinical decision support system CDSS form a significant part of the field of medical knowledge management technologies through their capacity to support the clinical process and use of knowledge, from diagnosis and investigation through treatment and long-term care. Their role and acceptance in daily clinical practice is increasing [15]. Recent studies [16] showed that CDSS can improve physician performance and accuracy, while the quality of each system may depend on the technical approach they use to model medical information. CPG are structured documents that list a set of specific cases for which some diagnosis and therapeutic action plans are recommended with a given degree reflecting the strength of the recommendation corresponding to the level of scientific evidence. CPG include the description of the various 844