The presently proposed research is intended to develop a prototype expert system that would assist tumor registrars in the performance of their case finding and abstraction duties. The user interface to this system would be a windows-based, mouse-controlled graphical user interface with features to make it easy to use for both the novice and seasoned registrar. The facilities provided by this system would complement those of existing commercial tumor registry software, which is primarily intended to automate registry operations once case finding and abstracting decisions have been made by the registrar. The expert system will process data (in electronic form) that are the sources of registry information. The expert system will incorporate probabilistic reasoning to account for uncertainties in the source data, and to fuse conclusions arrived at by independent paths of reasoning. The knowledge base of the expert system will be derived from two sources: the knowledge of experienced tumor registrars obtained during knowledge-engineering sessions, and the various published standards and guidelines which govern registry operations. The prototype is expected to serve as the basis for development of a comprehensive registrar's assistant, with possible additional features of quality control and registrar training.
