Development of Rule-based Expert System for Interpretative Report with Health Screening Tests

Lee, Chae Hoon

Abstract

BACKGROUND: Interpretative reporting is an important aspect of laboratory medicine. The large menu of laboratory tests available today makes it increasingly difficult for the non-specialist to order and interpret all laboratory tests. The aim of this study was to determine the usefulness of an expert system to interpret laboratory tests and help physicians order the appropriate tests.
MATERIALS AND METHODS
In order to interpret laboratory tests, a rules-based expert system was developed. In this module, if-then rules were used to interpret the given test result patterns (e.g. urinalysis, anemia, hepatitis B virus, hypercholesterolemia, glucose, syphilis, and tumor markers) and select matching text elements. The system was used to evaluate 535 subjects who visited a health-check program.
RESULTS
The overall abnormal rate was 50.5% in the expert system; 34% for cholesterol, 9.9% for urinalysis, 8.0% for anemia, 7.7% for thyroid function tests, 4.5% for tumor marker study, 4.7% for hepatitis virus antigen, 4.3% for serum glucose, and 1.1% for syphilis.
CONCLUSION
These results indicate that the application of the expert system for the interpretation of laboratory tests may provide a useful method for the interpretation of reports. However more rules are needed for the application to in-patients.