An intelligent information system for interpreting ECG tracings used in a CCU nurse training program

The expert/information system shell SMR (simulation of medical reasoning) was used to record ECG (electrocardiographic) morphology and interval data and to provide diagnostic interpretations and advice for therapy. The knowledge base contains decision rules relating the ECG data to 65 supraventricular arrhythmias. A treatment protocol is given for each disorder. An SMR decision rule is an inference statement and Boolean combination of criteria statements, all in natural language. Reasoning chains are formed with previous inferences used as criteria in subsequent rules. Mutually exclusive criteria referring to a numerical measurement or a qualitative fact are defined as criteria groups. Criteria statements and group names are organized into data screens for recording data. The nurse reads 25 ECGs and enters the findings into the system. The SMR report lists the data, the ECG diagnosis, and the suggested therapy. If the SMR conclusions differ from the trainee's, the SMR explanation facility is used to find the reasons for the discrepancy. The case record is updated and rerun until agreement is reached.<>