Computer assisted psychiatric diagnosis: experiments in software design

The process of psychiatric diagnosis involves the real-time utilization of large amounts of knowledge to maintain and test multiple hypotheses. Therefore, the development of an on-line computerized assistant that can aid a clinician performing psychiatric diagnoses presents challenging problems in data-base organization and retrieval. We have developed the DUNE (Diagnostic Understanding of Natural Events) system architecture that organizes the knowledge around processing structures. The system was designed as a shell for expert-systems that aid diagnoses and assessment tasks in ill-structured domains in general. These domains have previously proven unfeasible for traditional expert-systems such as rule based systems. Currently, DUNE contains sufficient knowledge to aid the diagnosis of anxiety and affective disorders. Among DUNE's advantages: a large degree of tolerance to clinician errors, and flexibility at run time. In this paper, we discuss the general specifications for a program that can aid psychiatric diagnoses, and then describe DUNE and its capabilities.