Hipokrat-I: Bayesian network based medical desicion support system

Medical Decision Support Systems are the software programs that help cliniciansduring the diagnostic process. These systems help clinicians to diagnose the patients under thegiven individual patient data. Medical Decision Support Systems bring a great value to qualityof patient care by early diagnosis, preventing errors, appropriate treatment. By this way costsare cut in the field of medicine. Since the field of medicine is very large this kind of systemscan handle only a subpart of the medical domain.Medical Syetems are started to be constructed by 1970?s. They had a over the toppopularity parallel to the Artifical Intelligence trend in 1980?s. After that, this topic had lostits popularity. But nowadays with the rapid development of technology (especiallydevelopment of the powerful hig speed processors and huge memory spaces), decisionsupport in the medical field become one of the interesting research area for academicians andsoftware developpers. Artifical Neural Networks, Fuzzy Logic, Rule Based Methods andBayesian Networks can be used to develop medical decision systems.In this thesis, methods mentioned above are explained and a decision support system isdevelop by using Bayesian Networks. This system covers the domain of thyroid diseases. Itcan diagnose different kind of thyroid illness. To construct such a system some electronicpatient records and traditionally kept(on the text) patient records are used. Help of a domainexpert is taken to determine the tests to be applied, consultations to be made and which of thesymptoms and findings effect the diagnosis process much.By this thesis a sub module of a hybrid diagnostic system is developped. This hybridsystem will combine a rule based module and bayesian network module. Both systems willhave their own inference mechanisms, results taken from both modules will be combinedlinearly under given weights then these results will be given to users.Keywords: Decision, Decision Support Systems, Medical Decision Support, ArtificalIntelligence, Rule Based System, Bayesian Network