Knowledge Base Development Environment and J2EE-Compliant Inference Engine for Clinical Decision Support

摘要

In order to provide high quality health care at low cost, it is essential that medical reasoning, decision making, and clinical action be directly supported by hospital information systems. We have developed a knowledge base technology to support the intellectual activities of medical professionals, and have applied it to clinical decision support in a hospital information system. In the case of placing an order, the hospital information system is able to provide summaries of clinical data, propose the appropriate parameters of an order, and validate contraindications within diseases, drugs, and laboratory results. This paper describes an overview of the hospital information system, the development environment of a medical knowledge base for clinical decision support, and its runtime environment. The development environment provides a knowledge representation language based on attribute grammar and debugging aids by using proof tree, so that medical professionals can audit and maintain a knowledge base by themselves. The runtime environment provides inference service through a stateless session bean of Java 2 Enterprise Edition, which wraps an inference engine built on the top of a multi-threaded Prolog. Experiments show that contraindications on average medication orders for a patient with non-severe illnesses can be checked within a practical execution time, and the inference service can be provided stably throughout load tests and long-term tests.