Design, implementation, use, and preliminary evaluation of a standards-based Web service for clinical decision support

Background. The healthcare system faces a care quality crisis. The provision of clinical decision support (CDS) can significantly improve care quality; however, most healthcare settings lack robust CDS capabilities. An important reason for this limited CDS availability is the difficulty of encoding medical knowledge in a machine-executable format that can be shared across institutions to enable CDS. To address this barrier to CDS adoption, a number of groups have developed approaches to sharing machine-executable medical knowledge and then integrating that knowledge into clinical applications to enable CDS. However, these approaches to knowledge sharing and integration are limited by the need to understand complex knowledge representation formalisms, the lack of support for different knowledge representation approaches, sparse content availability, limited implementation flexibility, the need for significant investments in infrastructure, and/or high implementation costs. Objective. To propose, implement, and validate an approach to knowledge sharing and integration that overcomes many of the limitations of existing approaches and allows machine-executable medical knowledge to be more easily re-used across applications and institutions. Methods. A knowledge sharing and integration approach based on the principles of a Service-Oriented Architecture (SOA) was conceived as a potential solution to the limitations of existing approaches. Then, the core component of this SOA approach---a service that uses patient data as the input and returns patient-specific conclusions as the output---was designed and implemented as a Web service known as SEBASTIAN. After completing a systematic review of the literature to identify the functional requirements of an effective CDS intervention, SEBASTIAN was used to implement four distinct CDS applications in a manner conformant with the results of the systematic review and with the principles of a SOA. Results. Preliminary assessments indicate that SEBASTIAN meets all original design objectives and that a SOA approach to knowledge sharing and integration can overcome the limitations of existing approaches. Conclusions. SEBASTIAN provides strong preliminary evidence that a SOA approach to knowledge sharing and integration could allow machine-executable medical knowledge to be more easily re-used across applications and institutions. Of note, SEBASTIAN is currently serving as the basis of the HU Decision Support Service standard.