In the context of the modern OR, medical and technical innovations for surgical assistance systems and process automation will be established. Therefore, Workflow Management Systems (WfMS) will be implemented in the operating room environment. Clinical and surgical processes as well as medical devices will be managed and monitored by such a process-driven software system . Hence, the description and visualization of the underlying surgical processes and activities in a machine readable format as SPMs is an obligatory requirement . Different aspects of surgical intervention such as surgical activities and behavior, actors, instruments and medical devices, materials as well as anatomical structures have to be taken into account for modeling surgical workflows . Unfortunately, models created by different process modelers will vary even if they use the same modeling language and modeling objective. Thus, the models often vary in aspects of granularity, naming and representation of process elements like tasks and activities, as well as in terms of modeling structure and layout. Especially the naming of process elements in natural language and thereby the usage of synonyms or homonyms is subject to a high level of variability . Furthermore, technical business process modelers and medical experts may not speak the same language and use different terminologies, so they have a different understanding and interpretation of the process models. The diversity in process modeling prevents an adequate representation and organization of process- and workflow models as well as their reusability for future modeling.
Hence, surgical workflow modeling requires a standardized methodological approach for generation, naming, visualization, storage and the search of process models. The semantic workflow modeling, which combines process modeling with terminologies, nomenclatures or ontologies, was presented as a solution for standardized workflow modeling in business processes [3, 4]. In the medical domain different vocabularies, thesauri, terminologies and ontologies for representing medical and surgical knowledge are available. Nevertheless, standardized medical terminologies, like SNOMED Clinical Terms (SNOMED CT), UMLS, LOINC or MeSH, have never been used for surgical process modeling. The aim of this paper is to present a methodology for a standardized semantic workflow modeling in the surgical domain. Additionally, a detailed concept of a framework for semantic workflow modeling, execution and management is presented.
2.1 Requirements to medical terminologies and ontologies
Semantic process modeling requires a formal ontology in combination with business process modeling . Thus, terminologies and ontologies can be used for formal representation of medical and surgical knowledge in a machine readable format. An ontology provides further information on a designated term as well as semantic relations between those terms. Figure 1 depicts common medical terminologies, ordered with respect to their level of formal semantics and standardization.
In order to select a suitable ontology for surgical semantic process modeling, different terminologies were analyzed. Hereby, functional, logical, organizational, informational and operational process representations as well as adequate resource representation for entities like instruments or medical devices have to be considered. Within a requirements analysis functionalities and properties of different medical terminologies were compared with each other. Thus, the requirements for an adequate surgical process description and modeling must be covered by the chosen terminology.
2.2 Requirements to surgical process modeling
Modeling surgical processes or workflows is used to describe and visualize operating room processes, existing procedures, activities, resources, tasks, participants, instruments, medical devices and anatomical structures. The express goal is to create an illustration of the actual processes and desired target processes inside the OR. First off, the functional process information, like inherent activities or tasks and their sequence of completion, are modeled. Depending on which point of view is adopted when creating the process model, different aspects, elements and information, like human and technical resources, have to be integrated in the model.
In business process modeling there is a large number of different modeling languages, which can be used for surgical process modeling as well. Process modeling languages based on XML, event-driven process chains or petri-net based modeling languages are most commonly used. For semantic annotation of process models, several business process modeling languages were analyzed with respect to their applicability for surgical semantic process modeling and their fitting to the medical ontology. Therefore, requirements analysis for adequate representation of surgical processes and process description in combination with the ontology was performed.
Based on the requirements analysis for surgical semantic process modeling, a system concept for a semantic process modeling and management framework was designed (see Figure 3).
3.1 Ontology mapping
As a result of the requirements analysis to medical ontologies, SNOMED CT was identified as a suitable ontology for the semantic representation of surgical knowledge. SNOMED CT is the most widely used nomenclature in medicine and includes over 800.000 terms, which describe 300.000 concepts and 1 million relationships between those concepts. SNOMED CT concepts represent clinical meanings, so medical activities, procedures and findings, symptoms as well as diagnoses, events, medical devices, instruments and materials can be described in a standardized manner. Compared to other medical ontologies, SNOMED CT provides a high level of formal semantics and standardization  and offers a hierarchical category system, which allows a structural classification of concepts (see Figure 2). Each concept has a unique identifier and is individually represented by a concept description. SNOMED CT concepts can be annotated by relational attributes, so hierarchical, temporal, local or methodological relationships between concepts can be described.
For surgical process modeling BPMN 2.0, a standardized and wide spread process and workflow modeling language, is well suited for surgical process and workflow modeling. Different approaches in literature present a semantic annotation for BPMN models . In addition, various approaches for combining medical process modeling with proprietary ontologies or knowledge bases exist . Nevertheless, process modeling languages like BPMN have not been annotated with commonly used or standardized medical terminologies and ontologies for surgical applications so far. Therefore, a concept for mapping BPMN with a standardized medical ontology (SNOMED CT) has been defined (see Figure 2).
First off, the semantical description of process model elements through the medical ontology requires a formal definition of the modeling language constructs, which could be mapped to the concepts and classes of the ontology. Both the formal semantics of the whole meta-process and the semantics of single process elements have to be represented by the ontological concepts and terms.
In general, process modeling languages consist of flow objects, like activity or event elements. Flow objects can be mapped to the case specific concepts and terms of SNOMED CT, for example procedure, event or situation concepts. Additionally, flow objects can be described in more detail with attached attributes. These attributes were mapped to SNOMED CT resource concepts like body structures, persons or devices. Every process element obtained a unique ontological identifier, a description, related synonyms and relational attributes. In addition, a logical and functional sequence between the flow objects is realized with the help of connectors such as sequence flow elements or split and join gateways. Connectors were mapped to the properties or relational attributes of ontological concepts, so that temporal, resource related, methodological or hierarchical aspects can be applied to the process model. SNOMED CT concepts are ordered hierarchically in different granularity levels, which can be applied to the SPM. Thereby, a homogenous granularity level within a SPM can be achieved. Hierarchical linkages between process models in different granularity levels were realized through subprocess or transaction elements, which were mapped to the hierarchical ordered concepts of the ontology.
Consequently, the elements of the ontology and the process modeling language can be mapped to each other in a 1:1 relationship. Therefore, formal transformation rules and appropriate mapping algorithms have to be defined for every ontological concept and process modeling element.
3.2 System concept
For a standardized semantic workflow modeling of surgical processes a desktop modeling tool was designed, which allows BPMN process modeling with SNOMED CT ontology integration. Based on SNOMED CT concepts and the ontology inherent knowledge, the modeling tool enables context sensitive suggestions for the naming and structural modeling of the actual process step. Besides functionalities and properties for surgical process and workflow modeling with BPMN, the tool provides an annotation component for ontology integration. Therefore, SNOMED CT concept identifiers and descriptions were integrated in the representation of flow objects and process attributes. Hence, a standardized naming and multilingual description as well as a homogenous granularity level within the process model through the SNOMED CT hierarchical ordered concepts can be achieved. By the usage of ontology inherent relational attributes, process knowledge in the form of subprocesses, pre and post conditions for the current process step as well as required methods and materials can be derived and automatically integrated in the SPM. Adequate search functions for descriptions or synonyms and filter rules will be provided by the annotation tool. In addition, there is a test and update component, which allows update management for the underlying ontology and an automatic replacement of process fragments. By using semantical and syntactical test methods, a verification of model correctness and model consistency could be reached. The SNOMED CT concepts are stored in the ontological repository, which is accessed by the modeling tool. Furthermore, a workflow repository was conceptualized, which allows organization and hierarchical representation of the semantically annotated work-flows in different granularities. In the third repository semantical rules, like guidelines or business rules, are organized. The modeling tool provides a rules editor for creating and editing those rules.
In the context of the execution environment an ontological reasoner enables the orchestration of atomic work-flows to provide executable (meta) workflows through semantic web services. The reasoner operates on base of the probabilities for process execution, defined business rules and workflow models stored in the workflow repository. The emergent workflow models will be initialized, executed, managed and monitored by a workflow engine. Therefore, the WfMS provides a web based administration and execution tool as user interface. An essential part of the framework is the monitoring component, which logs data about workflow orchestration and execution. The recorded probabilities for performed workflow orchestration are stored in the rules repository and therefore can be used for better prediction of future orchestration.
With ontology integration in SPMs, a standardized naming through ontological concepts and identifiers, a homogenous granularity and structural modeling will be achieved. The automatic creation and completion of process models could be provided by the intended semantic modeling tool, so the variability and working effort for surgical workflow modeling can be reduced. Based on inherent ontological knowledge, process models could be enriched with relational attributes, inherent methods, guideline-based processes, pre and post conditions or time aspects for automatic completion of process modeling steps. These semantic annotations enable an enhanced search in ontological concepts and descriptions with synonyms as well as the validation and reuse of semantic workflow models. The executable semantic workflows will be orchestrated by an ontological reasoner based on semantic web services for workflow management.
In this paper a methodology and a framework for standardized semantic workflow modeling in the surgical domain, was presented. The system aims to support an appropriate, functional and standardized workflow modeling and execution, so that an adequate workflow management in operating room environment can be realized.
ICCAS is funded by the German Federal Ministry of Education and Research (BMBF) and the Saxon Ministry of Science and Fine Arts (SMWK) in the scope of the Unternehmen Region with grant number 03Z1LN12 and by the European Regional Development Fund (ERDF) and the state of Saxony within the frame of measures to support the technology sector.
Franke S, Liebmann P, Neumuth T. Connecting workflow management to the OR network: Design and evaluation of a bridge to enabe dynamic systems behaviour. In: Biomed. Eng. Biomed. Tech., vol. 57, no. SI-1 Track-N, 2012: pp. 771–774 Google Scholar
Neumuth T, Jannin P, Strauss G, Meixensberger J, Burgert O. Validation of Knowledge Acquisition for Surgical Process Models. In:J. Am. Med. Inform. Assoc. JAMIA, vol. 16, no. 1, 2009: pp. 72–80 Google Scholar
Ehrig M, Koschmider A, Oberweis A. Measuring Similarity Between Semantic Business Process Models. In:Proceedings of the Fourth Asia-Pacific Conference on Conceptual Modelling . Vol. 67, Darlinghurst. 2007: pp. 71–80. Google Scholar
Thomas O, Fellmann M. Semantic Process Modeling – Design and Implementation of an Ontology-based Representation of Business Processes. In:Bus. Inf. Syst. Eng., vol. 1, no. 6, 2009: pp. 438–451 Google Scholar
Jimeno-Yepes A, Jiménez-Ruiz E, Berlanga-Llavori R, RebholzSchuhmann D. Reuse of terminological resources for eflcient ontological engineering in Life Sciences. In:BMC Bioinformatics, vol. 10, no. Suppl 10, 2009: pp. 1–13 Google Scholar
Abramowicz W, Filipowska A, Kaczmarek M, Kaczmarek T. Department of Information Systems, Pozna [nacute] University of Economics. Semantically enhanced Business Process Modelling Notation. 2007 Google Scholar
Burgert O, Neumuth T, Lempp F, Mudunuri R, Meixensberger J, Strauß G, Dietz A, Jannin P, Lemke HU. Linking Top-Level Ontologies and Surgical Workflows. In:Int J CARS, vol. 1, 2006: pp. 437–438. Google Scholar
About the article
Published Online: 2015-09-12
Published in Print: 2015-09-01
Conflict of interest: Authors state no conflict of interest. Material and Methods: Informed consent: Informed consent is not applicable. Ethical approval: The conducted research is not related to either human or animals use.