| Popis: |
Matching appropriate computational techniques to particular knowledge domains is a major problem for information technology research. A class of computational techniques currently of considerable interest is the Rule-based system. One successful rule-based system is MYCIN, a consultation system to advise physicians on the use of antibiotics. We are investigating whether the essential structure of MYCIN is transferable to another medical domain, clinical psychopharmacology, which differs from antibiotic therapy in significant ways in the kind of knowledge used in the system. Our investigation is based on an evaluation of the level of performance attained by the system and on an analysis of the effort required to incorporate different kinds of knowledge into the system. In our work, we have discovered what we believe is an essential design problem for medical expert systems: controlling the amount and the type of information which a system requests from a user. This problem is inherent in medical expert systems because of the nature of the distribution of clinical states and of the training and background of physicians. The problem also exists for human consultants, and a complete and general solution for computer systems is probably not achievable. Our experience also suggests that undue concern over conventional distinctions among system architectures may be unrewarding and that progress in developing medical expert systems is strongly dependent on the quality of software tools. |
