Evolution of a knowledge base for a clinical decision support system encoded in the Arden Syntax

Area of Science:

• Medical Informatics
• Computer Science in Medicine

Background:

• Clinical decision support systems (CDSS) are increasingly integrated into medical practice.
• The long-term viability of CDSS relies heavily on the effective maintenance of their knowledge bases (KB).

Purpose of the Study:

• To quantify the nature and frequency of changes within a real-world CDSS knowledge base over an extended period.
• To analyze the evolution of Medical Logic Modules (MLMs) written in Arden Syntax.

Main Methods:

• Retrospective analysis of 156 Medical Logic Modules (MLMs) from the Columbia-Presbyterian Medical Center's CDSS KB.
• Examined 2020 distinct MLM versions developed over 78 months, detailing 5528 changed statements.

Main Results:

• Significant evolution of the CDSS KB was observed, with 5528 statement changes across 2020 MLM versions.
• Changes were concentrated in specific areas: logic (38.7%), action (17.8%), queries (15.0%), and data slots (12.4%).

Conclusions:

• Long-term maintenance of CDSS knowledge bases necessitates substantial and ongoing modifications.
• Findings have implications for designing more effective KB editors for Arden Syntax and similar systems.