Ontology-based modelling of stroke clinical pathways

Abstract

Healthcare spending in Canada is on the rise. One method to reduce healthcare spending is to reduce length of stay (LOS). Clinical Pathways (CPs) are one recommended management technique to reduce LOS. CPs are implementations of medical guidelines in a specific healthcare environment. This may include hospitals, clinics or other healthcare facilities. They represent an evidence-based patient care workflow for a specific disease. The adoption of CPs allows for easier continuity of care across different healthcare settings and medical teams. While the use of CP as part of standard patient care has grown considerably in the past decades, not much progress was made in CP representation and modeling to encode CP data properly within existing Health Information Systems (HIS). One proposed method to achieve this goal is ontological modeling. Ontology is a formal model that represents a certain subject matter. It not only communicates what things exist in a certain domain or field but also how those things relate to each other. This research proposes an ontological model for stroke CP representation and processing. Such a model would allow CPs to be sharable, extendable, and machine-readable, thus enabling greater patient management. The Systematized Nomenclature of Medicine – Clinical Terms (SNOMED – CT) is used to encode medical knowledge described within clinical pathways. The stroke CP Ontology is an extension of a generic CP ontology, with new concepts introduced specific to the domain of stroke. It is able to represent different types of CP activities, occurring over a period of type, referencing medical knowledge contained in SNOMED CT. It is also able to infer new knowledge using the Semantic Web Rule Language (SWRL). This ontology is presented to users through a prototype Clinical Pathway Management System (CPMS). The CPMS is built using Java and the Eclipse IDE. The OWL and SWRL API are used to directly connect to and query ontology files. After completion of a CP, the CPMS generates new ontology files unique to each patient’s CP execution as well as a general output file of patient activities and outcomes. Data analytics can be performed on this output file to determine the most common CP activities, levels of compliancy and similarities between patient CP progressions.