Test-selection Strategies for Probabilistic Networks

摘要: Decision-support systems are used in a large variety of domains. In the medical domain, such can be equipped with patient-specific facility that indicates which diagnostic test or combination tests should performed. Current systems, however, do not take into account might ordered packages rather than one by one. furthermore physicians would prefer to gather information for specific purpose, as general condition patient, prior gathering about an other topic, instance presence absence metastases. this thesis, we describe automatic test-selection designed fits better daily practice current facilities do. Our is capable selecting while taking characteristics patients illness, test, its sensitivity and specificity predictive value, order various different subgoals investigated physicians. It able select multiple just test. By extending decision-support system facility, become more interesting use practice. A good could lead ordering fewer tests. This imply lower financial costs, shorter periods waiting visit physician receive therapy. Futhermore, it patient care higher quality general.