Title :
The fuzzy receiver operating characteristic function and medical decisions with uncertainty
Author :
DeLeo, James M. ; Campbell, Gregory
Author_Institution :
Div. of Comput. Res. & Technol., Nat. Inst. of Health, Bethesda, MD, USA
Abstract :
Receiver operating characteristic (ROC) methodology is an important tool in decision making where there are two outcomes. By introducing the notion of partial degree of membership, fuzzy ROC methodology extends classical ROC methodology as fuzzy set theory extends classical two-state set theory. This paper introduced fuzzy ROC methodology and demonstrates its importance in decision making. Computational methods are developed and important concepts are demonstrated with examples from clinical medicine. These concepts include objective derivation of fuzzy functions with linguistics modifiers, enhanced certainty in sequential case learning, fuzzy inferencing using multiple fuzzy variables and associations with artificial intelligence expert systems and neural networks
Keywords :
decision theory; expert systems; fuzzy set theory; inference mechanisms; medical diagnostic computing; neural nets; artificial intelligence expert systems; clinical medicine; decision making; fuzzy functions; fuzzy inferencing; fuzzy receiver operating characteristic function; fuzzy set theory; linguistics modifiers; medical decisions; medical diagnostic computing; neural networks; partial degree of membership; sequential case learning; uncertainty; Artificial intelligence; Artificial neural networks; Decision making; Fuzzy neural networks; Fuzzy set theory; Fuzzy systems; Hybrid intelligent systems; Learning; Medical expert systems; Set theory;
Conference_Titel :
Uncertainty Modeling and Analysis, 1990. Proceedings., First International Symposium on
Conference_Location :
College Park, MD
Print_ISBN :
0-8186-2107-9
DOI :
10.1109/ISUMA.1990.151339