Research On Knowledge-based Semantic Modeling And Variance Handling Mechanism For Clinical Pathway Workflows

The healthcare system is an important research and application domain of the industrial engineeringdiscipline. The service quality and efficiency of healthcare delivery directly affect public benefits andsocial stabilization. Clinical pathways are an important approach to increasing healthcare efficiency andquality and decreasing healthcare costs. They define the optimal multidisciplinary care processperformed by a team of health care professionals for a particular diagnosis or procedure and have beenwidely recognized and adopted by the world’s healthcare systems as a care-process-orientedmanagement pattern. Currently, they are mostly managed and monitored in a paper-based manual way,whose inherent deficiencies significantly impair total effectiveness of clinical pathwayimplementations.Computerized modeling, execution and monitoring of clinical pathways have become a keyapproach to increasing their efficiency and realizing efficient healthcare services. Considering thecharacteristics of much knowledge involvement, complex variances and their fuzzy handling in clinicalpathway management, this dissertation studies the following two key problems for implementingclinical pathway computerization from the viewpoint of knowledge engineering and management aswell as implementation technology:(1) knowledge semantics modeling of clinical pathway workflows;(2) variance handling support and management.On the basis of introducing the concept of semantics-based workflow management, this dissertationproposes an ontology-based framework of clinical pathway workflow and variance management tofacilitate the realization of semantics management of clinical pathway workflows. Main contributionsof this dissertation are as follows:A clinical pathway ontology (CPO) is presented and constructed on the basis of extending theprocess ontology in OWL web ontology language for services (OWL-S), i.e., extended processontology (EPO), to provide a common semantic foundation for knowledge representations of allclinical pathways. First, this dissertation extends the process ontology in OWL-S to construct the EPOmodel. It is used to support the descriptions of resources, organization structures and temporal knowledge required by the executions of service processes and serves as a formal semantic foundationfor modeling the knowledge related to the execution of general processes during workflowmanagement, which therefore helps to overcome the problem of not describing the semantics of processand temporal knowledge in general workflow models, especially those constructed using XML-basedworkflow language. Secondly, on the basis of EPO, this dissertation identifies and formally defines inOWL web ontology language (OWL) basic concepts and relations in general clinical pathways to buildthe CPO model. It not only provides explicit semantic foundation for describing any specific clinicalpathway knowledge, but also facilitates consistent exchange, sharing and interoperability of relevantknowledge semantics between clinical pathway workflow management and healthcare professionals,other applications and systems (e.g., HIS).The CPO-based semantic modeling methodology of clinical pathway workflows is proposed foraddressing the descriptions of knowledge semantics in the clinical pathway domain and provideing ageneral method of knowledge modeling for various clinical pathways to realize computerizedscheduling and monitoring of patient care workflows based on the knowledge semantics of clinicalpathways. The methodology includes a hierarchical modeling approach and a modeling approach totemporal knowledge based on rules in semantic web rule language (SWRL). The former describesclinical pathways at two interconnected semantic levels, namely outcome flow level and interventionworkflow level. It enables reusable semantic representations of standardized healthcare workflowthrough different abstract layers and structured semantic description components and provides flexiblesupport for modeling semantic changes of clinical pathways due to variance handling. The latterrepresents time interval relations between clinical interventions during the same or different care days.The combination of the approach and temporal entities in the CPO model can sufficiently describevarious temporal knowledge semantics in clinical pathway workflows. In addition, a case study of aclinical pathway for cesarean section demonstrates the applicability of the proposed methodology inenabling explicit, structured semantic models of real clinical pathway for other diseases or proceduresand thus providing an important foundation fro the realization of semantics-based clinical pathwayworkflow management.A hybrid variance handling approach based on the generalized fuzzy event-condition-action(GFECA) rule and typed fuzzy Petri net extended by process knowledge (TFPN-PK) is proposed byintroducing fuzzy set theory and methods for addressing the problem of variance handling support inclinical pathway workflows. It enables integrated representation and reasoning of fuzzy/non-fuzzyknowledge and supports two strategies of direct decision and analysis-based decision for handling different types of variances to facilitate dynamic management of clinical-pathway-based patient careworkflows. GFECA rules are used to support the handling of variance events related topatients/families, medical staffs and healthcare resources by specifying fuzzy/non-fuzzy events andconditions, classical actions and event/condition thresholds. The TFPN-PK model and proposed fuzzyreasoning algorithm are used to support the analysis and handling decision of most clinical varianceevents. The model, as an extended FPN, not only models fuzzy relationships between fuzzypropositions, but also builds semantic associations among fuzzy propositions, domain ontologyconcepts in the healthcare domain and process knowledge to enable integrated modeling of domainontology and knowledge and workflow processes. The TFPN-PK-based weighted fuzzy reasoningalgorithm addresses the problem of fuzzy reasoning involving uncertain goal propositions and knowngoal concepts by combing foreword reasoning and backward reasoning, which in turn facilitates theinteraction and integration between variance knowledge reasoning and clinical pathway workflowapplication. Finally, this dissertation explains the architecture, analysis and design models of a variancemanagement system in detail and successfully develops its prototype. It also uses the prototype tosupport variance handling during the implementation of the clinical pathway for cesarean section,which shows that the proposed approaches and algorithms are effective for fuzzy variance analysis andhandling.The proposed models, approaches and algorithms in this dissertation provide sufficient, formalknowledge semantics descriptions for clinical pathway workflows and support the decisions onanalyzing and handling different types of variance events, which provide an important foundation forrealizing computerized implementation and dynamic management of semantics-based clinical pathwayworkflows. Basic solutions and approaches in this dissertation provide important reference value forsemantic modeling, execution, and control of workflows and exception management in other domains(e.g., manufacturing).