Abstract
The SAGE (Sharable Active Guideline Environment) Project is multi-site, interdisciplinary, research and development effort to enable encoding and broad dissemination of medical knowledge in the form of computable clinical practice guidelines. The vision of the SAGE Project is that, once encoded, guideline content could be deployed to, and used to provide clinical decision support via the native functions of many, heterogeneous clinical information system platforms. During the five-year project, IDX (now part of GE Healthcare) worked in partnership with Apelon Inc., Intermountain Health Care, Mayo Clinic, Stanford Medical Informatics, and the University Of Nebraska Medical Center to achieve all major project objectives. In this session, we identify key innovations in each of the main project focus areas: The SAGE Guideline Representation Model, the Protégé-based guideline encoding workbench, the SAGE Guideline Execution Engine, and the standards-based interfaces with host clinical information systems. In each area we also describe how our consortium worked collaboratively to define requirements, resolve technical and informatics challenges, and validate prototypes in end-to-end testing.
DESCRIPTION AND OUTLINE OF PRESENTATION
-
Overview of SAGE Project Vision, Objectives, and Collaborative Approach.
-
1.1. Project objectives and approach
The SAGE Project Vision: An innovative, standards-based infrastructure for representing and deploying medical knowledge in the form of computable clinical practice guidelines.
Key project objectives: A standards-based guideline representation model; a guideline authoring/encoding workbench; a guideline execution engine; a methodology for downloading and localizing encoded guidelines; and a standards-based infrastructure for interacting with host clinical information systems.
-
1.2 Overview of SAGE Project Collaborative Approach
An interdisciplinary team of clinicians, health informaticians, and system developers – formed across industry, academia, and centers of clinical excellence.
Combination of joint requirements-definition and problem-solving exercises, integrated with focused prototyping teams.
A requirements-driven methodology, using clinically valid exemplars of guideline usage as use cases in “end-to-end” (encoding through execution) prototyping and testing.
Outreach beyond project boundaries to align with (or drive) best available informatics standards and to implement national standard vocabulary resources for decision support.
-
-
How we collaborated to develop the SAGE Guideline Representation Model.
-
2.1. Innovative Features of the SAGE Guideline Model
Integrates standard medical vocabularies and information models.
Designed to model key aspects of healthcare workflow and define opportunities for decision support.
Combines “top-level” workflow activity graphs with detail-level encoding of medical decision logic and computable guideline recommended actions.
Employs a flexible and efficient formalism for representing and querying “evidence statements” – evidence-based associations between patient conditions and possible interventions.
-
2.2. Development of the SAGE Guideline Authoring/Encoding Workbench
Use of an enhanced Protégé platform as an open-source authoring and validation environment.
Integration of standards-based terminology services during the encoding process.
Generation of a document-oriented view of encoded guideline
-
-
How we addressed the challenges of encoding guideline content with computable specificity.
-
3.1. Collaborative development of the SAGE knowledge-engineering, and guideline encoding processes.
Disambiguation and organization of source guideline content.
Defining care context scenarios and opportunities for clinical decision support.
-
3.2. Use of standard medical terminologies during guideline encoding.
Mapping and modeling of medical and operational concepts.
Translating guideline recommendations to executable SAGE decision models.
-
-
How we employed standards-based terminology services and utilities.
-
4.1. Use of standard terminology services during both authoring/ encoding of guideline content.
Integrating terminology services with the guideline encoding application.
Integrating terminology services with guideline execution at runtime.
-
4.2. Using information models to extend standard medical vocabularies.
Use and extension of NCVHS standards (SNOMED-CT, LOINC, NDF-RT)
Runtime integration with external knowledge bases.
-
-
How we designed, developed, and tested an interoperable guideline deployment system.
-
5.1. Key features of the SAGE Guideline Execution Engine
Scalable, multi-threaded execution of guideline logic.
Processing of complex guideline recommendations (e.g., patient-specific order sets).
-
5.2. Standards-based interfacing with the host clinical information system.
Real-time detection of events in the clinical workflow.
Communicating with the host CIS using the VMR (virtual medical record) standard.
-
5.3. How we demonstrated interoperable deployment of computable guidelines to multiple platforms
Installation of SAGE-encoded guidelines to multiple platforms.
Surfacing guideline content via native CIS applications and screens.
-
-
How we solved problems of adapting guideline content to local environments.
-
6.1. Installation of executable guideline content at local care delivery organizations.
Supporting review of computable guideline content by non-technical clinicians.
Local editing of guideline content.
Mapping and binding to local terminologies and data services.
-
6.2. How to present active guideline recommendations to busy clinicians.
Integrating advanced guideline recommendations via the clinical workflow
-
Educational Goals
Attendees will understand the challenges of encoding and disseminating computable clinical guidelines.
Attendees will understand the vision and objectives of the SAGE Project.
Attendees will understand why a collaborative effort was vital to achieving project objectives.
Attendees learn of key SAGE Project innovations and how they were achieved.
Who Should Attend
Clinicians and informaticians with an interest in the use of advanced clinical decision support to improve health care quality, improve patient safety, and reduce health care costs.
Researchers interested in the nature of a large scale, complex R&D collaboration.
Presenters
Robert Abarbanel, MD, Ph.D., GE Healthcare, Seattle, WA.
James R Campbell, M.D., University of Nebraska Medical Center, Omaha, NE.
James Guy Mansfield, Ph.D., GE Healthcare, Seattle, WA.
Mark Nyman, M.D., Mayo Medical School, Mayo Clinic, Rochester, MN.
Samson Tu, MS, Stanford Medical Informatics, Stanford University, Stanford, CA.
Tony Weida, Ph.D., Apelon, Inc., Ridgefield, CT.