Verification & validation of the knowledge base for the hypertension management CDSS.

Abstract

To implement a knowledge-based clinical decision support system for clinical information systems, it is crucial to verify and validate the knowledge base. This study developed and tested the hypertension management CDSS, named LIGHT. This study used a knowledge representation framework based on SAGE and developed a knowledge converter to translate knowledge encoded into the knowledge engine. To verify knowledge converted through the knowledge converter that is included in the knowledge representation framework, expected recommendations were made according to the knowledge encoded based on 201 test cases. The expected recommendations were compared to those generated by the knowledge engine. To validate the knowledge base, two physicians reviewed the test cases and made medication orders according to the knowledge base. These medication orders were compared to recommendations generated by the LIGHT. The concordance rates for compelling indication and absolute contraindication were 85% and 100%, respectively. Another senior physician reviewed and analyzed the discrepancy cases between the orders of the two other physicians and system recommendations. Accordingly, the authors conclude that the knowledge base for hypertension management became more accurate and practical through the testing process.