A Domain Specific Expert System Model for Diagnostic Consultation in Psychiatry

Medical Experts Systems have been one of the earliest and ongoing pursuits of the Artificial Intelligence (AI) community. Unfortunately, they still remain a largely unrealized goal. The systems that have been developed are either prototypes, or involve only small knowledge domains. One of the main reasons for this situation is explained as the lack of domain specific models. While it is true that there are significant differences in the nature of the clinical knowledge and the diagnostic strategies used by expert clinicians across different medical subspecialties, this fact has been largely overlooked. In addition, Knowledge engineers who are not the domain experts have failed to capture the uniqueness, depth and the complexity of clinical reasoning. This has resulted in expert system models that are too generic and non-intuitive to clinicians. Psychiatry is characterized by its highly subjective and vague knowledge and reasoning process. Therefore generic models, which have not taken this into consideration, are particularly unsuitable for psychiatry. The authors have introduced an expert system model specific for psychiatry, in which diagnostic knowledge is described as a hierarchically organized set of entities through which diagnostic inference is made via a bottom-up approach. The relationships between the entities in diagnostic knowledge are described in terms of likelihoods and the degrees of severity using approximated mathematical functions. While this model highlights the need for domain specific models, it also gives insight to the implementation of similar approaches in other medical subspecialties. It is the intention that this model will be implemented as a web-based diagnostic consultation system.