Modeling the Arden Syntax for medical decisions in XML

https://doi.org/10.1016/j.ijmedinf.2008.01.001Get rights and content

Abstract

Purpose

A new model expressing Arden Syntax with the eXtensible Markup Language (XML) was developed to increase its portability.

Methods

Every example was manually parsed and reviewed until the schema and the style sheet were considered to be optimized. When the first schema was finished, several MLMs in Arden Syntax Markup Language (ArdenML) were validated against the schema. They were then transformed to HTML formats with the style sheet, during which they were compared to the original text version of their own MLM. When faults were found in the transformed MLM, the schema and/or style sheet was fixed. This cycle continued until all the examples were encoded into XML documents. The original MLMs were encoded in XML according to the proposed XML schema and reverse-parsed MLMs in ArdenML were checked using a public domain Arden Syntax checker.

Results

Two hundred seventy seven examples of MLMs were successfully transformed into XML documents using the model, and the reverse-parse yielded the original text version of MLMs. Two hundred sixty five of the 277 MLMs showed the same error patterns before and after transformation, and all 11 errors related to statement structure were resolved in XML version. The model uses two syntax checking mechanisms, first an XML validation process, and second, a syntax check using an XSL style sheet.

Conclusion

Now that we have a schema for ArdenML, we can also begin the development of style sheets for transformation ArdenML into other languages.

Introduction

The Arden Syntax is a language designed to represent the context of, support for and logic behind medical decisions and to do so in a syntax that renders these decision executable by computers. It is used to construct executable medical logic modules (MLMs) that represent individual clinical decisions. Arden Syntax version 1.0 was adopted as a standard by the American Society for Testing and Materials as document E 1460 in 1992. Later, version 2.0 was proposed and adopted as a standard for representing MLMs by HL7 in 1999 [1].

Arden Syntax is composed of maintenance, library and knowledge categories with appropriate slots. The maintenance category contains the slots that specify information unrelated to the health knowledge in the MLM. These slots are used for MLM knowledge base maintenance and change control. And they also contain information about the version of the Arden Syntax that is being used. The library category contains the slots pertinent to knowledge base maintenance that are related to the MLM's knowledge. These slots provide predefined explanatory information and links to facilitate searching through a knowledge base of MLMs. The knowledge category contains the slots that actually specify what the MLM does. These slots define the terms used in the MLM (data slot), the context in which the MLM should be evoked (evoke slot), the condition to be tested (logic slot), and the action to take should the condition be true (action slot).

Although Arden Syntax is the HL7 standard, there are problems related to the adoption of Arden Syntax as the standard tool for describing medical logic. Two prominent problems are: (1) the curly braces problem and (2) the compiler problem. The term “curly braces problem” comes from the Arden approach to mapping MLM variables to clinical data sources. The structure and terminology of stored clinical data often vary from institution to institution. The approach taken in Arden is to provide a location, separated from the rest of the MLM by curly braces, where queries, expressed in the query language of the local Electronic Health Record (EHR) can be expressed. Data collected in local hospital environments can then be mapped to MLM identifiers. In the future, XML-based techniques like the Virtual Medical Record (vMR) [1] as well as other techniques based on the HL7 Reference Information Model (RIM) [2] may provide solutions to this problem. In addition to the challenges associated with differing models for the storage and retrieval of stored data, different clinical systems rely on different software development environments. As a result, different Arden Syntax compilers have been necessary to fit MLMs for the specific architectures and information systems of each hospital [3].

Building compilers specific to the Arden Syntax, however, is difficult; several institutions have tried to compile Arden Syntax to C++ [4], [5], to Mumps source code [6], to JAVA [3] or to a pseudo-code compiler-interpreter [7], and then re-compile (or interpret) the source codes to create a functioning execution environment (Fig. 1). The key problem remains: specific compilers are required for hospitals using different computer languages.

XML is commonly recognized as a component of the worldwide web used to format content that is easily transformed to HTML pages. In addition to its characteristics as a markup language, XML is now an industry standard for data communication among different breeds of computer systems. During the past few years, XML has been introduced into the healthcare industry and is now being widely used. Applications of XML in health care have a wide range including academic studies that contain descriptions of the contents of heart sound components [8], representations of models of biochemical reaction networks [9], ADE reports [10], and pharmaceutical inquiries [11]. XML is also entering the medical domain through standards activities that include inter-system messaging (HL7 3.0), the clinical document architecture (CDA), as well as knowledge representations, including clinical guidelines [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] and the Arden Syntax.

XML can be used to convert clinical guidelines to electronic form [8], [15], [16], [19], [21], [22], [23], [24]. This conversion makes it easy to transform the guidelines to human-friendly form and to machine-friendly form, and makes it possible to share guidelines among medical computing groups. Easy transformation of any document to another form is one of the greatest benefits of XML. There has been considerable progress in the representation of clinical guidelines in XML, and several studies in this area are still being conducted by researchers. Other researchers, however, have mainly focused on the representation of expert knowledge [15], [16], [17], [20], [21], [23], [25], [26] in electronic form, or on the extraction of a medical logic module [21], [27] from those electronic forms.

Arden Syntax version 3.0 is currently under development by HL7. This version is expected to be based on XML as is the case with much of the next generation of HL7 standards. In this paper an Arden Syntax in XML, and MLMs written in this XML-based format, are called “ArdenML” and “MLMs in ArdenML” respectively. Sailors has described one goal of ArdenML as “translating MLMs into other program constructs’ knowledge representations without having to build a traditional compiler.” [24] According to his conceptualization, the challenge of creating an ArdenML can be divided into 5 levels, numbered 0–4 (Table 1). Development of a level-4 schema, including operands and operators, is necessary to meet the goal. Recent examples of ArdenML, however, remain at level 3 or have reached only the early stage of level 4. Continued development of xPath and XLST techniques are encouraging the development of ArdenML at the operators and operands level.

The objective of this study is to increase the portability of the Arden Syntax. To address this issue, we will build a XML schema representing this syntax and develop an XSL style sheet that can reverse-parse MLMs in “ArdenML (Arden Markup Language)” [24] to readable HTML, allowing easy comparison to the original text and thus verify MLMs written in this XML model. In this study, XML will be used to present MLM in a form that may substitute for the parse tree that is produced by compiler. A transformation of this XML was used to produce the original text version of MLM for validation purposes. This paper proposes newly developed Arden MLMs with XML format as a first step to address this compiler issue. In this form, the MLMs are, at least partially, precompiled.

Section snippets

Materials and tools

The Arden Syntax for Medical Logic Systems, version 2.1 [28] including a context-free grammar represented by the Backus-Naur Form (BNF) defines the current standard. A context model used in MathML from http://www.w3c.org/math was used in designing the expression language for ArdenML. The goal was to build a schema capable of representing a large group of existing MLMs and to demonstrate the success of this effort by converting these MLMs back to their ASCII form (HTML) using an XSL style sheet.

Results

All the schema, style sheet and examples in both text and ArdenML version are available in the Internet web site (http://61.78.109.24/ArdenML).

Discussion

In this work, we demonstrated one approach to the goal of expressing Arden-based medical logic as XML. Notably, a key challenge was finding a good model for the expression language embedded in Arden Syntax. We were able to find a stable, workable schema by introducing context model from MathML. RuleML [29], a general knowledge representations tool in the web, has also expression language of its own. We, however, took the advantage of MathML over RuleML because of several redundant tags in

Acknowledgement

This work was supported by the Post-doctoral Fellowship Program of Korea Science & Engineering Foundation (KOSEF).

References (32)

  • M. Hucka et al.

    The systems biology markup language (SBML): a medium for representation and exchange of biochemical network models

    Bioinformatics

    (2003)
  • S. Kataoka et al.

    Developing and integrating an adverse drug reaction reporting system with the hospital information system

    Yakugaku Zasshi

    (2002)
  • M. Dugas et al.

    Design and implementation of a common drug information database for a university hospital

    Pharm. World Sci.

    (2003)
  • T. Ganslandt et al.

    An XML-based system for the flexible classification and retrieval of clinical practice guidelines.

  • G. Georg et al.

    Interpretative framework of chronic disease management to guide textual guideline GEM-encoding

    Stud. Health Technol. Inform.

    (2003)
  • S. Hoelzer et al.

    Value of XML in the implementation of clinical practice guidelines--the issue of content retrieval and presentation

    Med. Inform. Internet Med.

    (2001)
  • Cited by (20)

    • Comprehensive analysis of rule formalisms to represent clinical guidelines: Selection criteria and case study on antibiotic clinical guidelines

      2020, Artificial Intelligence in Medicine
      Citation Excerpt :

      A pharmacogenetic CDSS that integrates clinical and genomic data employed Fuzzy Arden and a Probabilistic OWL reasoner in [50]; Lyme borreliosis CIGs and many clinical surveillance and Web systems are also Arden-based [52–55,58]. Owing to the lack of Arden free execution environments, some researchers have opted to transform Arden MLMs into XML [45], Arden ML for further rule serialization with Drools or other formats [51,59]. A glimpse of the future direction and evolution of the Arden Syntax is provided in [57], while in [24], the authors defend Arden’s strengths, such as its temporal operators.

    • Arden Syntax: Then, now, and in the future

      2018, Artificial Intelligence in Medicine
    • Transformation of Arden Syntax's medical logic modules into ArdenML for a business rules management system

      2018, Artificial Intelligence in Medicine
      Citation Excerpt :

      However, Blaze Advisor requires explicit variable types and declaration statements before use. To meet the syntax of SRL, an optional attribute called ‘otype’ was used to make any implicit data type explicit for its transformation to other languages [11]. In addition, the XSLT stylesheet was designed to collect all variables in the logic and action slots of the MLM, remove duplicates, and put them at the beginning of the SRL with a declaration statement.

    • Accessing complex patient data from Arden Syntax Medical Logic Modules

      2018, Artificial Intelligence in Medicine
      Citation Excerpt :

      Clinical demands suggest that population-based MLMs will be of growing interest in the near future and will demand an approach for mapping complex data. Recently (November 2014) ArdenML [26,27] was integrated into the latest version 2.10 of the standard, enabling entire MLMs, including statements and operators, to be expressed in XML. A standardized XML or JSON data format for the deserialization approach would be another meaningful extension to the standard as it would complement the existing READ and READ AS statements with a convenient way to map random data structures of random complexity onto Arden Syntax data types.

    • Multi-model-based interactive authoring environment for creating shareable medical knowledge

      2017, Computer Methods and Programs in Biomedicine
      Citation Excerpt :

      Another intrinsic barrier for creating MLMs is the curly brace problem of querying the required input data from the medical systems. The CDSS community has recommended a standard information model, virtual medical record (vMR), to resolve the issue of heterogeneous data models for CDSS systems [18] as well as the curly brace problem [19]. The standard data model vMR meets the scalability and interoperability objectives for a knowledge acquisition tool [18,20].

    View all citing articles on Scopus
    View full text