URUZ
Introduction- what are clinical guidelines?
Clinical guidelines (or Care Plans) are a powerful method for standardization and
uniform improvement of the quality of medical care. Clinical guidelines are a set of
schematic plans, at varying levels of abstraction and detail, for management of
patients who have a particular clinical condition (e.g., insulin-dependent diabetes).
(Clinical protocols are typically highly detailed guidelines, often used in areas such as
oncology and experimental clinical trials.) The application of clinical guidelines by
care providers typically involves collecting and interpreting considerable amounts of
data over time, applying standard therapeutic or diagnostic plans in an episodic
fashion, and revising those plans when necessary.
Motivation for developing Uruz
Most clinical guidelines are in free text and are inaccessible to the physicians who
most need them. Even when guidelines exist in an accessible electronic format,
physicians rarely have the time and means to decide Which guideline best pertains to
their patient, and how to apply that guideline to Their particular patient.
Thus, there is an urgent need to facilitate guideline dissemination and application
using automated methods, and the Thus, the question is how to structure the large set
of existing free-text clinical guidelines to support effective search, retrieval, and
browsing, as well as application and quality assessment.
The Conversion Problem
The core of the guideline conversion problem is that (1) expert physicians cannot
(and do not need to) program in a guideline specification language, while knowledge
engineers do not necessarily understand the clinical semantics of the guidelines; (2)
text-based representations are useful for search and retrieval, while formal
representations are essential for automated execution.
Therefore, the authoring tool for clinical guidelines will use a three-level hybrid
representation: Free-text, semi-structured-text and machine-comprehensible, fullystructured text (formal-language)
The DeGeL hybrid-representation model
To gradually convert a large mass of free-text clinical guidelines to a set of target
ontologies, we have developed a Web-based, distributed architecture, the Digital
electronic Guideline Library, DeGeL),and several web-based tools, which gravitate a
guideline gracefully from text-based, through structured text (labeled by the
knowledge roles of a target ontology), to a fully formal, machine- readable,
executable representation (Figure 1).
Free-text
guideline
Adding a machineSemantic markup
(semi-Structuring) comprihensible format
Web-based
guideline library
a
b
My
line
guide
-Eligibility & applicability determination
-Runtime application
-Retrospective Quality assessment
Context-sensitive
Web-based Search,
retrieval, and
Browsing interface
Figure 1. The hybrid meta-ontology and incremental conversion process in the DeGeL architecture. Input freetext
guidelines are uploaded from various sources and loaded into a markup editor, in which expert physicians
classify and structure the free-text guidelines using the knowledge roles of target guideline ontology. Knowledge
engineers add executable expressions by filling additional levels in the target-ontology’s representation.
representation without a full computational model.
Uruz - the DeGeL library’s Web-based clinical guideline markup tool
The Uruz Web-based guideline markup tool (Figure 1) enables medical experts to:
create new guideline documents. A source guideline is uploaded into the DeGeL, and
can then be used by Uruz to create a new guideline document, marked-up by the
semantic labels of one of the target ontologies available in DeGeL. Uruz can also be
used to create a guideline document de-novo (i.e., without using any source) by
directly writing into the knowledge roles of a selected target ontology. We are
developing an Asbru-dedicated tool to add the formal-specification level.
Figures 2 and 3 show the Uruz semantic-markup interface. The user browses the
source guideline in one window, and a knowledge role from the target ontology in the
other window. She labels the source content (text, tables, or figures) by dragging it
into the knowledge-role frame. Note that the editor can modify the contents or add
new content. This enables turning implicit knowledge into more explicit, further
facilitating the task of the knowledge engineer who fully formalizes the guideline.
Fig.2. The Uruz Web-based guideline markup tool. The tool’s basic interface is uniform across al guideline ontologies.
The target ontology selected by the medical expert, in this case, Asbru, is displayed in the upper left tree; the guideline
source is opened in the upper right frame. The expert physician highlights a portion of the source text (including tables or
figures) and drags it for further modification into the bottom frame labeled by a semantic role chosen from the target
ontology (here, filter condition). Note that contents can be aggregated from different locations in the source. The bottom
left textbox, Element Comments, stores remarks on the current selected knowledge-role, thus supporting collaboration
among guideline editors.
A more complex module embedded in Uruz, the only one specific to the Asbru
ontology (such modules can be defined for other ontologies), the plan-body wizard
(PBW), is used for defining the guideline’s control structure (see Figure 4). The
PBW enables a user to decompose the actions embodied in the guideline into atomic
actions and other sub-guidelines, and to define the control structure relating them
(e.g., sequential, parallel, repeated application). The PBW, used by medical experts,
significantly facilitates the final formal specification by the knowledge engineer.
When a knowledge engineer needs to add a formal, executable expression to a
knowledge role, she uses one of the ontology-specific Uruz modules (we are
developing one specific to Asbru), which delves deeper into the syntax of the target
ontology.
For example, in our hybrid Asbru, conditions can include temporal patterns in an
expressive time-oriented query language used by all of the application modules. To be
truly sharable, guidelines need to be represented in a standardized fashion. Thus,
Uruz enables the user to embed in the guideline document terms originating from
standard vocabularies, such as ICD-9-CM for diagnosis codes, CPT-4 for procedure
codes, and LOINC-3 for observations and laboratory tests. In each case, the user
selects a term when needed, through a uniform, hierarchical search interface to our
Web-based vocabulary server.
Fig. 3. The Asbru plan-body wizard (PBW) module. On the left, the guideline’s structure tree
is displayed and updated dynamically as the user decomposes the guideline. On the upper
right, the user is prompted with wizard-like questions to further specify the selected control
structure. In the bottom right, the text of the source, current, or parent guidelines is displayed.
Future Work
Currently work is done towards URUZ Ver.3 which will :
• Graphical and intuitive WinForm tool for guideline structuring
• Enable structuring for semi and formal language
• every element in the formal-language will have special graphical
representation , emphasizing its special characteristics(e.g plan-body
builder,condition builder)
• Support multi-ontology for mark-up
• Part of DeGeL’s framework
• the tool will also involve interaction with other tools (IDAN’s frame work)