Medical decision making research has tended to emphasize the generation of optimal decisions, an issue which is central to the development of clinically useful consultation programs. This paper stresses the need to consider other theoretical and practical issues that are pertinent if consultation systems are to be accepted by physicians. Since adequate decision making performance remains an essential component of acceptable systems, the paper suggests c-iteria for selecting clinical problems that may be amenable to short-term implementation using state-of-the-art techniques. Introducticn At the beginning of a third decade of research into the development of computer-based diagnostic aids, it is appropriate for medical computer scientists to assess the strides that have been taken, the barriers that remain, and the optimal strategies for furthering the field in the years ahead. One purpose of this meeting is to take a thoughtful look at medical decision making research and to identify potential solutions to the theoretical and logistical problems that continue to abound [1],[2].

These diverse roles impose several requirements consultation programs has become known as »expert systems upon the system.

Data structures arc described to GEV using the GLISP Structure Description language III; if the user is already programming in GLISP.

This paper describes techniques for representing control knowledge in preaicate calculus. A hybrid system is described in which metarules for diagnostic problem solving and their interpreter (both supplied by the NEOMYCIN program) are expressed in a form of predicate calculus (supplied by MRS). Procedural attachment is used to access and execute the untranslated domain knowledge. A simple deliberation/action loop manages the system at the highest level. There are three metalevels of reasoning.

The Stanford University component of this research is funded in part by ARPA contract #MDA903-80-C-0107, NIH contract # NIH RR 00785-10, ONR contract #N00014-79-C-0302. Compiled oy Bruce G. Buchanan November 1982 Abbreviations Used in This Bibliography: AAAI American Association for An:ficial Intelligence ACM Association for Computing Machinery AFIPS American Federation of Information Processing Societies ECAI European Conference on Artificial Intelligence IEEE Institute for Electrical and Electronic Engineers IFIPS International Federation of Information Processing Societies IJCAI International Joint Cr nferences on Artificial Intelligence SIGPLAN ACM Specia! Abe, N., ltoh, F., and Tsuji, S. Toward a learning of object models using analogical objects and verbal instruction. Addis, T. R., and Hartley, R. T. A faultfinding aid u,sing a content addressable file store. ICL Technical Note TN 79, ICL Ltd., London, 1979.

The first computers were numerical calculators that did not appear to embody much intelligence at all. But before these machines were even designed.

In this paper I take an empirical look at the question of whether there are rational memckis of discovery and claim that computer programs provida a laboratory for experimentation on this question Recent work in artificial intelligence or Al. has produced programs capaole of serious intellectual work in science Results from Al,viii be used to show that there exist mechanized procedures for discw.ering

Artificial intelligence, long a topic of basic computer science research, is now being applied to problems of scientific, technical, and commercial interest. Some consultation programs, though limited in versatility, have achieved levels of performance rivaling those of human experts. A collateral benefit of this work is the systematization of previously unformalized knowledge in areas such as medical diagnosis and geo!ogy.

List of Tables Table 4-1: Status determined based on the ''alues returned by selection rules 13 ACKNOWLEDGMENTS 1 would like to thank

The application of Artificial Intelligence techniques to real-world problems has produced promising research results, but seldom has a system becoma a useful tool in its domain of expertise.