This paper describes an informal but systematic method for how to test and verify a knowledge-based system in a large open-ended medical target domain. The system used is Guardian, an intelligent system for monitoring and diagnosis of post-cardiac surgery patients in an intensive-care unit. The knowledge base is tested and verified by running the system on a series of realistic test scenarios, both with an embedded simulator and with an external simulation system. The same scenarios are presented to human test subjects, making it possible to compare and analyze the performance of the knowledge-based system with that of human physicians. The use of simulators instead of clinical data also means that it is possible to test crucial scenarios which occur seldom in medical practice. Our results show that a system like Guardian might indeed be useful in medical care.

Describes the use of explicit models of goals and functions for monitoring and diagnosis of intensive-care patients. The method is based on multilevel flow models (MFM) and used in the Guardian system. It provides this system with alarm analysis, fault diagnosis, and automatic generation of explanations. Advantages include a relatively easy knowledge engineering effort and good properties for use in a system with hard real time deadlines. The results of some experiments are also reported.