This paper presents a wiki environment for the modelling of Computer Interpretable Guidelines (CIGs) using the graphical language DiaFlux. We describe a wiki-driven development process using a stepwise formalization and allowing for almost self-acquisition by the domain specialists. The applicability of the approach is demonstrated by a project developing a guideline for sepsis diagnosis and treatment by a collaboration of clinicians.
In this paper, we introduce the graphical language DiaFlux
for modeling of clinical guidelines: The contributions are
its simple application for developing decision-support
systems, only providing a limited number of intuitive language
elements. Due to its simplicity, it is possible to be used
by domain specialists and thus eases the application in the
knowledge engineering process. To allow for comfortable
development of DiaFlux guidelines, we introduce a visual
editor integrated into the Semantic Wiki KnowWE [
Clinical guidelines have shown their bene ts by providing
standardized treatment based on evidence-based medicine.
Many textual guidelines are readily available and also shared
through the internet, but rely on the proper application by
the clinician during the actual care process. Much e ort
has been put into the development of formal models for
computer-interpretable guidelines. In the variety of CIG
models, each has its own focus, e.g. the sharability of
guidelines between various institutions or assisting patient care
through active decision-support [
For the speci cation of a clinical guideline, two types of
knowledge have to be e ectively combined, namely
declarative and procedural knowledge. While the declarative part
encompasses the facts and their relationships, the
procedural one re ects the knowledge about how to perform a task,
i.e. deciding which action to take next. DiaFlux guidelines
are based on owcharts. Distinct node types allow, e.g.,
the conduction of tests, the changing of diagnosis states and
calling other DiaFlux guideline as self-contained modules.
To express the procedural aspect of the guideline, nodes are
connected by edges, which are labeled with conditions. They
evaluate the declarative knowledge with respect to the
observed ndings, e.g. the outcome of a given test or the status
of a diagnosis. To obtain the semantics necessary for
executability, we rely on an application ontology as an
extension to the task ontology of diagnostic problem solving [
Collaborative development requires to track the changes of all participants. Therefore, a frequent task is to compare di erent versions of a wiki article. For DiaFlux guidelines a visual di is provided, highlighting changes for easy comparison. After creating a knowledge base in the wiki, a test session can directly be started from the wiki. The current state of the guideline throughout the session can be observed, highlighting the traversed pathway through the guideline. This immediate feedback considerably eases the interactive testing of the knowledge base, especially when parallelism or hierarchical structuring are involved. For the development of DiaFlux models we propose a stepwise formalization process: At rst, informal information are collected in wiki articles, e.g. about goals of a protocol. During the next step, a rst semi-formal guideline is created using only nodes containing free text and connecting edges. At this stage of formalization, the owcharts can not be automatically executed, but \manually". For testing purposes the user can explicitly select the pathway through the guideline. The taken pathway is highlighted for easier tracking. The last step is the full formalization into a DiaFlux model and the creation of the application ontology, resulting in an executable knowledge base. By following this process of gradual re nement, the entry barrier for domain specialists is quite low, while knowledge acquisition can start from the beginning.
In the context of the project \CliWE" we used a prototype of the clinical wiki environment for the development of a guideline covering the diagnosis and therapy of sepsis. Sepsis is a syndrome of a systemic in ammation of the whole body with a high mortality (30 to 60%).
The knowledge base was developed in accordance to the o cial guideline by the German Sepsis Society. Our formalization of the guideline covers so far the diagnostics and parts of the therapy (cf. Figure 1). It contains about 50 nodes in eight modules with several possible pathways, depending on how the diagnosis can exactly be established and the course of the therapy. Overall, the wiki-based approach showed its applicability and usefulness, as the combination of formal and informal knowledge and its gradual re nement was intensely used during the acquisition of the guideline and according test cases.
We thank Prof. Dr. Norbert Weiler and Dr. Dirk Schadler, University Medical Center Schleswig-Holstein, Campus Kiel, for their e orts during the development of the sepsis guideline.