Galapagos: Scalable Medical Terminology Development

In the complex and ever-expanding world of medical informatics, the development and maintenance of large-scale medical terminologies present significant challenges. As terminologies like SNOMED International grow in size and complexity, developers are often limited by the lack of effective tools to manage a collaborative and distributed development process. Traditional methods often fail to scale cognitively, computationally, and organizationally, hindering progress and innovation.

This seminal paper, “Scalable Methodologies for Distributed Development of Logic-Based Convergent Medical Terminology,” introduces a groundbreaking solution: the Galapagos environment. This system addresses the critical need for a development environment that supports concurrent, multi-site enhancements to complex, logic-based medical terminologies.

Key Innovations and Concepts:

• Evolutionary vs. Creationist Design: The authors challenge the traditional “creationist” design philosophy—where a pre-ordained, singular design is rigidly implemented—arguing it is stifling for large-scale projects. Instead, they advocate for an “evolutionary design” philosophy. This approach embraces heterogeneity, encourages diverse approaches, and allows the terminology to evolve and adapt through a continuous development and feedback process to meet the needs of a broader group of users.
• Overcoming the “Local-Update Penalty”: A major obstacle in terminology development is the “local-update penalty,” where users who make the most significant local enhancements face the greatest difficulty reconciling their work with new official releases. Galapagos is designed to reverse this penalty by providing tools that assist in upgrading local terminologies and ensuring local enhancements are reflected in new reference versions.
• Semantics-Based Concurrency Control: Traditional concurrency control methods, such as locking mechanisms, are unsuitable for the long-lived transactions typical of terminology development. The paper proposes a semantics-based concurrency control approach. Instead of relying on strict transaction serialization, this method uses semantic information and Aristotelian classification principles (genus and differentia) to identify and resolve conflicts between concurrently developed enhancements.
• The Galapagos Environment: Built upon the K-Rep knowledge-representation system, Galapagos is a configuration management and conflict resolution environment. Its key features include:
  • Conflict Detection: It algorithmically detects two primary types of semantic conflicts: the multiply-defined term conflict (one term, multiple definitions) and the non-unique definition conflict (different terms, same definition).
  • Interactive Conflict Resolution: Recognizing that resolving conflicts requires domain knowledge, Galapagos provides tools to facilitate resolution by terminology modelers. It supports merging definitions, verifying synonymy, and disseminating changes efficiently.
  • Configuration Management: The system uses a change-set model, an extension of the long-transaction model, which allows changes to be captured as independent, named entities. These change sets can then be exchanged and applied to different configurations, supporting distributed development without centralized coordination.

Proven Performance in a Real-World Setting:

The paper presents a “proof-of-performance” evaluation conducted over a six-month period, involving concurrent enhancement of SNOMED International at three Kaiser Permanente regions and the Mayo Clinic. The evaluation demonstrated that:

• The system effectively identified and managed conflicts, with an overall conflict rate of just 2.2% over more than 54,000 transactions.
• The conflict resolution process was manageable, typically taking an afternoon, which is a small price for enabling parallel development.
• Critically, the process of discussing and resolving conflicts prompted meaningful discussions among modelers, leading to an improved shared conceptual model and a deeper understanding of the modeling task.

This research provides a robust framework and a proven toolset for collaborative, scalable, and evolutionary development of the complex medical terminologies that are foundational to modern electronic medical records and health informatics applications.

Reference:

Campbell, K. E., Cohn, S. P., Chute, C. G., Shortliffe, E. H., & Rennels, G. (1998). Scalable methodologies for distributed development of logic-based convergent medical terminology. Methods of Information in Medicine, 37(4-5), 426–439.