Using intermediate states to improve the ability of the Arden Syntax to implement care plans and reuse knowledge.

E H Sherman, George Hripcsak, Justin Starren, Robert A. Jenders, Paul D. ClaytonPubMed 1995

TL;DR

This paper introduces a novel approach to enhance the Arden Syntax, a clinical decision support language, by storing patient states as "Intermediate States" in a central data repository. This method significantly improves the handling of complex, time-dependent care plans, facilitates knowledge reuse across different applications, and allows for modular development even when data is incomplete.

What Problem Does This Paper Address?

Automating clinical practice guidelines and care plans holds great promise for improving patient outcomes. However, existing knowledge representation languages, including the Arden Syntax, struggle with the temporal complexity inherent in many medical protocols. Specifically, representing and tracking sequences of events over time, sharing and reusing knowledge effectively, and dealing with missing patient data have been significant hurdles. This limits the widespread adoption of automated clinical decision support systems in everyday practice.

Methods

The core innovation presented is the adaptation of the clinical event monitor to store coded data representing "Intermediate States" of a patient within a care plan. Instead of MLMs (Medical Logic Modules) directly calling each other, they now interact by reading and writing these Intermediate States to a central data repository. This shifts the paradigm from a tightly coupled network of MLMs to a more modular, data-driven approach.

Key aspects of the method include:

  • Intermediate States: Coded data representing the current status of a patient within a care plan (e.g., "New Warfarin Rx," "Diet I Rx suggested").
  • Central Data Repository: A database that stores these Intermediate States, along with a timestamp.
  • Modular Development: MLMs are designed to be developed and tested independently, triggered by specific data or Intermediate States.
  • Data-Driven Interaction: MLMs communicate and influence each other indirectly through the shared Intermediate States, rather than direct calls.

Datasets

Not discussed in the paper. The paper focuses on the methodology and its application to four constructed care plans.

Key Insights and Ideas

  • Decoupling for Temporal Complexity: Storing Intermediate States decouples MLMs, making it much easier to manage and track complex sequences of events over extended periods. This avoids the need for intricate, brittle MLM networks.
  • Enhanced Knowledge Reuse: Intermediate States serve as a valuable source for knowledge reuse. They can be used by other MLMs within the same care plan, by unrelated MLMs in the broader knowledge base, and even by non-decision support applications.
  • Modular Development Freedom: The ability to develop and test MLMs independently, without pre-determining the entire care plan structure, significantly simplifies the knowledge engineering process. This allows for piecemeal development and easier adaptation.
  • Handling Unobtainable Data: This modular approach allows for the implementation and testing of parts of a care plan even when preceding data is missing. This is crucial in real-world scenarios where complete patient data is rarely available.
  • Bridging the Gap to Usability: By addressing these core limitations, the approach makes automated care plans more practical and adaptable, paving the way for wider adoption in clinical settings.

Key Results

The paper demonstrates improved ability to:

  • Track complex series of events and recommendations over long periods.
  • Reuse generated knowledge by the originating care plan, other knowledge base elements, and non-decision support applications.
  • Deal with unavailable data by enabling modular implementation of care plan components for which sufficient data exists.

The authors constructed four care plans to evaluate these benefits, with specific examples like the Warfarin therapy and Cholesterol care plans illustrating the practical advantages.

Limitations and Future Work

  • Knowledge Sharing vs. Reuse: While knowledge reuse is significantly enhanced, the paper notes that knowledge sharing (transferring MLMs between institutions) is not directly improved by this method.
  • Dependencies: While modularity is a key benefit, the authors acknowledge that direct dependencies between some decisions in a care plan still exist, meaning not all parts can be developed independently in all cases.
  • Future Work: The authors suggest that the Intermediate States could be used by other applications, such as populating patient problem lists or providing information to healthcare provider workstations. They anticipate further exploration of these diverse applications.

Summary updated

Abstract

The Arden Syntax is one of a few knowledge representation languages currently in use for clinical decision support. While some of these languages are being used in active patient care settings, none have gained widespread acceptance as a clinical tool. Prior attempts to represent temporally complex care plans in the Arden Syntax have revealed difficulties in representing and tracking series of consecutive time-oriented events and recommendations, in sharing and reusing knowledge and in dealing with unobtainable data. In an attempt to improve Arden's ability to deal with these problems and demonstrate the importance of these factors, the clinical event monitor has been adapted to store coded data representing Intermediate States in the Columbia Presbyterian Medical Center (CPMC) central data repository. The Intermediate States define the current state of the patient as laid out in the care plan. Four care plans were constructed. The findings include an improved ability to track complex series of events and recommendations over long periods of time. The knowledge generated by the electronic care plans was able to be reused by the care plan that generated it, by other elements of the knowledge base and by non-decision support applications. Modular development, facilitated by the changes, simplified dealing with data not available to the central data repository by aiding the implementation of those parts of the care plan for which sufficient data is available.
BibTeX
@inproceedings{sherman1995using,
  author = {E H Sherman and George Hripcsak and Justin Starren and Robert A. Jenders and Paul D. Clayton},
  title = {{Using intermediate states to improve the ability of the Arden Syntax to implement care plans and reuse knowledge.}},
  booktitle = {PubMed},
  year = {1995},
}
Figure