The design of a rule-based clinical event monitor in a multi-vendor hospital computing environment.

The Clinical Event Monitor (CEM) described here is a prototype system designed to explore the issues involved in building an institutional CEM that permits rapid, automated evaluation of clinical transactions and notification to clinicians of exceptional events in a multi-vendor computing environment. The CEM uses expert systems, database, and systems integration techniques. Ancillary (departmental) applications, including as Patient Registration, Laboratories, and Pharmacy have been licensed from commercial vendors. Application-to-application and application-to-database interfaces were built to mirror subsets of the ancillary patient databases into an institutional relational database (Oracle). The CEM receives registration updates via an HL7 message and evaluates data dependencies in rules via an interface to the relational database. The CEM engine was built using Nexpert, a commercially available expert system shell. Our short term goals were to: (1) build and maintain a patient census within the expert system environment via net based HL7 update broadcasts; (2) explore the data-driven features of Nexpert, (3) deliver prototype exception reports. This paper describes in general terms the design features of the CEM and in detail the features of a patient registry to NEXPERT bridge (from Oracle via HL7 structured transactions to NEXPERT) and the delivery of exception reports.