Background

last modified:
01/30/2017 - 21:50

Below is the BRIDG Project timeline that explains the evolution of the project since it's inception.

2003—2004: CDISC and HL7 RCRIM WG

The idea of building a Domain Analysis Model (DAM) which documented the shared semantics of the protocol-driven-research domain began with a September, 2003 dialogue between the CDISC Board of Directors and Charlie Mead,  The dialogue was, in turn, in response to a request from CDISC leadership to HL7 leadership that the two organizations renew a then 2-year-old Memorandum of Understanding (MoU) to work collaboratively on data exchange standards in domains deemed to be of interest to both organizations. In the context of the renewal of the MoU, CDISC requested HL7 provide an educational session for the CDISC BoD that would provide them with a better understanding of HL7's model-based approach to message specification, a process that CDISC had attempted to embrace in the context of the HL7 RCRIM Work Group with somewhat limited success. 

Following a discussion of the benefits, risks, and effort involved in pursuing the path of building a DAM, the CDISC BoD elected to begin funding the development of a CDISC DAM. The process officially started with a scope definition meeting in March, 2004, the resulting DAM ultimately becoming the BRIDG Model approximately one year later.  The current scope definition of the BRIDG domain is essentially the same as was defined at this initial meeting, i.e. protocol-driven research plus its associated regulatory artifacts.
 

2004—2005: The NCI caBIG® (and CDISC) Structured Protocol Representation Teams

In February 2004, NCI began a project to develop a structured model of a clinical trial protocol as part of the caBIG® project, awarding a grant to the University of Pittsburgh Medical Center (UPMC) to form the Cancer Structured Protocol Representation Team to focus on identifying a set of data elements suitable for a computational representation of structured protocols. Rather than create "yet another standard," this Team began to seek out other groups interested in standardizing the exchange of information related to clinical trials research with whom they might collaborate. The search naturally led to interactions with representatives from both CDISC and the HL7 RCRIM WG. An outgrowth of this collaboration was the inclusion of the caSPR Team—and thus the NCI and caBIG®—in the building of the still-to-be-named "CDISC Domain Analysis Model project."  The first joint modeling sessions involving both members of the CDISC DAM team and the NCI team were held in the late fall of 2004 (approximately 9 months after the start of the still-to-be-named BRIDG Project), with the meetings continuing over the next 9 months (when the project officially adapted the name "BRIDG" (Biomedical Research Integrated Domain Group), a name chosen more for the meaning of the acronym than the collective meaning of the words behind it). This initial effort, the combination of the work that CDISC had done during the first 9 months of CDISC Domain Analysis Model Project and the first attempts to integrate the work of the NCI team, resulted in the first Domain Analysis Model of the domain of protocol-driven research, which at that time was referred to by many as the "CDISC Problem Space Model."

The efforts of the following key individuals were essential to the early success of BRIDG:

Landen Bain (CDISC)
Ken Buetow, Ph.D. (NCI)
Julie Evans (CDISC)
Doug Fridsma, M.D., Ph.D (University of Pittsburgh)
Smita Hastak (ScenPro, Inc.)
Becky Kush, Ph.D. (CDISC)
Charlie Mead, M.D., M.S. (Booz Allen Hamilton, Inc./NCI)
 

2005—2006: Maturation of the BRIDG Project

The project was officially renamed "BRIDG" in the summer of 2005, in no small part to draw attention to the bridging of the semantic gaps that normally impede semantic interoperability (SI) in general and Computable Semantic Interoperability (CSI in particular. In addition, there was considerable motivation to find an as-yet-unused name for the project to prevent the confusing adoption of organization-specific monikers, e.g. CDISC Problem Space Model, HL7 RCRIM Domain Analysis Model, NCI Model. The acronym "BRIDG", which, as mentioned above, technically stands for "Biomedical Research Integrated Domain Group," was chosen to reflect the project's contribution to building both a variety of different "bridges" between the various stakeholder organizations. 

With the inclusion of multiple stakeholders, it became apparent that a more formal organizational governance structure must be defined and put in place if the effort was going to succeed in the long-term. Several of the individuals involved in these initial BRIDG project efforts felt that the project could productively take guidance on consensus building and management of diverse, distributed content from various standards-building organizations such as HL7, W3C, and OASIS. Likewise, experience on the management and evolution of shared content could be drawn from the open-source community in the form of projects such as Linux, Firefox, and Mozilla.

In concert with beginning to articulate the governance of the BRIDG Project, the ever-expanding model was ported from Rational Rose to Sparx Systems Enterprise Architect (EA) tool in the Spring of 2005, in part because of the higher cost without higher (relevant) functionality of the former application, but in larger part to allow the dissemination of the model to a larger community using the free model reader available from Sparx Systems, the makers of EA (www.Sparxsystems.com). In addition, the model itself was moved from residence on a single laptop to a GForge site to take advantage of a number of GForge capabilities such as version control, document management, and bug tracking. 

An early version of the BRIDG Model was presented to the HL7 RCRIM WG in late-2005/early-2006 and was subsequently accepted as the single Domain Analysis Model for use by the HL7 RCRIM WG in support of their compliance with the HL7 Development Framework (HDF) and its increased focus on robust representation of implementation-independent requirements. This commitment meant that not only would the HL7 RCRIM WG use the BRIDG Model going forward for all of its message specification efforts, but that it would also commit to reverse engineer all of the implementation-independent semantics from existing RCRIM WG specifications (e.g. Regulated Product Submission, Structured Product Labeling, etc.) into the BRIDG Model, a commitment which is in part evident in the content of the Regulated Product Submission specification in Release 1.0 of the BRIDG Model. At approximately the same time, the CDISC BoD committed to reverse engineering all existing CDISC specifications (e.g. SDTM, ODM, etc.) into the BRIDG Model as well as to adopt a formal technical specification process for future CDISC standards that included the utilization of the BRIDG Model. All four of the BRIDG stakeholder organizations are currently involved with applying "BRIDG in context" in a number of application and/or data interchange projects focused on computable semantic interoperability. 

Amidst these activities, it became clear to the stakeholders of the project that additional strategic oversight was needed within the BRIDG Project. As a result, the BRIDG Board of Directors (BoD) was created in August 2005 with representation from each of the four BRIDG Project stakeholder organizations, i.e. HL7, NCI, CDISC, and the FDA. The BRIDG BoD was charged with identifying, securing, and allocating resources to support the modeling effort, establishing priorities for modeling and harmonization efforts, and managing the vetting and review of the evolving model within their stakeholder organizations. 

Similar to other open source or standards development organizations as mentioned above, it was also decided that a "Technical Harmonization Committee" (THC) (now called the "Semantic Coordination Committee" (SCC)) needed to be established to manage the operational details of the model construction, harmonization, maintenance, and dissemination. The SCC was tasked with defining and executing the harmonization activities that would result in the representation of project-specific semantics in the larger context of the complete BRIDG Model, i.e. in assembling the various building blocks of the BRIDG Model into a coherent form as they were incrementally presented over time by various stakeholder project teams. The SCC's responsibilities were defined/restricted to making representational rather than semantic changes in the overall BRIDG Model, as well as to ensure that ambiguous semantics across the various projects contributing to the BRIDG Model were effectively normalized (different terms for the same concept mapped to that concept) or harmonized (same term for different concepts disambiguated through the creation of an appropriate number of newly-named concepts). Like the BRIDG BoD, the SCC has representatives from the four stakeholder organizations with specific modeling and/or associated domain-centric informatics skills.  Both the SCC and the BRIDG BoD are discussed in more detail below.
 

2007 - 2012: BRIDG Evolution

As BRIDG became more visible in the community, the BRIDG SCC started hearing feedback from both domain experts and technologists. Many domain experts said that the model was too large, too technical, and did not represent their semantics in language they could understand. The technologists said that the model was not clear enough to map cleanly to downstream models (e.g., design models, implementation models, etc.). As a result, the BRIDG BoD and SCC designed a BRIDG solution that would present each audience of BRIDG with its own perspective. 

The BRIDG Model uses a multi-perspective approach to modeling the static semantics of protocol-driven research. Each perspective tailors the representation of the semantics to a different audience. To learn more about the different BRIDG perspectives, please see the documentation at http://bridgmodel.nci.nih.gov/What/technical-representation.

In addition, the BRIDG BoD decided in late 2008 to move BRIDG forward as an international standard through the ISO. CDISC took this forward to ISO. ISO assigned the identifier “14199” to BRIDG and Technical Committee (TC) 215 approved the recommendation to circulate BRIDG as a DIS ballot: During the summer of 2010, the 254 comments from that ISO Joint Initiative Council (JIC) ballot cycle (along with a parallel comment period within CDISC and a ballot within HL7) were processed and included in the next two releases of BRIDG,

There were multiple BRIDG releases during this time period. To see the entire list of BRIDG releases and the projects that were harmonized, please go to What's in BRIDG tab on this website.

2013 – 2016: BRIDG Scope Expansion – Translational Research

Following a gap in the BRIDG project in late 2012 to early 2013, during which the BRIDG BoD was reorganized into the BRIDG Advisory Panel, the leadership of BRIDG determined that an NCI model called the Life Sciences Domain Analysis Model (LS DAM), a peer of BRIDG that had leveraged common concepts already defined in BRIDG, should be harmonized with BRIDG. The resulting model would then span the full range of basic, pre-clinical, clinical and translational research. This process started in 2013 and completed in 2015 (there was a gap in the project from late 2013 to mid-2014). The LS DAM harmonization introduced 3 new sub-domains covering concepts related to biospecimens, molecular biology and experiments.

In 2014, a new work group called the HL7 BRIDG Work Group was started. This group meets on a regular basis (every other week) to discuss issues and progress on the BRIDG model. The HL7 BRIDG work group provides a platform for BRIDG users and implementers to discuss BRIDG from an implementation viewpoint and identify areas of enhancement. That same year, the BRIDG Advisory Panel was reorganized into the BRIDG Steering Committee which governs the development and maintenance of the model still performed by the SCC.

Subsequently, in early 2015, the CDISC Pharmacogenetic and Pharmacogenomic (PGx) domains were harmonized with BRIDG to complement the related genetic/genomic concepts added to the model for LS DAM. The PGx concepts leveraged and helped flesh out some concepts identified earlier by the LS DAM harmonization.

With the release of BRIDG 4.0, the scope of BRIDG is now officially expressed as “basic, pre-clinical, clinical, and translational research and associated regulatory artifacts”. 4.0 is the first release of BRIDG as a translational research model and therefore some of the newer sub-domains are not fully fleshed out as yet. BRIDG 4.0 was balloted in May 2015 in the HL7 community as a Draft Standard for Trial Use (DSTU) with ballot comments coming from the FDA and clinical genomics communities. Subsequent resolutions of the comments have resulted in model changes, documentation updates and a new architectural focus on making BRIDG more SME-friendly.

Additionally, many comments highlighted the fact that the scope of BRIDG now encompasses sub-domains for which we don’t have ready access to SMEs but which are yet highly relevant to integration with the broad translational research scope BRIDG is trying to address. Consequently, the new architectural focus of BRIDG is about trying to work out how BRIDG might model such specialized semantics as clinical genomics and imaging by referencing them in external models that already contain the semantics BRIDG users need to represent. This “modeling by reference” approach means that BRIDG will be working closely with other industry-established standards, such as DICOM, to harmonize requirements, determine boundaries and address the mechanics of the process.

2017 – Present: Imaging and Modeling by Reference

The US National Cancer Institute (NCI) has use cases that need integration and harmonization of semantics between the clinical research domain and the Imaging domain. BRIDG Release 5.0 includes the new Imaging sub-domain to address these use cases.

The focus was specifically to harmonize relevant parts of the DICOM standard and a NCI project – Annotated Imaging Markup (AIM). NCI’s AIM project has created a standard means of adding information and knowledge to an image in a clinical environment, so that image content can be easily and automatically searched.

The scope of this Imaging harmonization was limited to support the following high level use cases:

  • Identification of entities – person, animal, specimen, image
  • Image acquisition
  • Image Type (modalities – CT, MR and PET)
  • Annotation & Structured Reporting

One of the key aspects of this harmonization of imaging semantics is that the BRIDG modeling team has leveraged the principles of “modeling-by-reference” which essentially means that when an established standard exists in a particular domain (like DICOM for Imaging) then not all the semantics of the referenced standard will be harmonized with BRIDG. Instead the harmonization effort will focus on aligning the common semantics between the two domains to support implementable interoperability use cases. The focus was to identify the touch points between BRIDG and DICOM to support users finding sufficient commonly used, high-level imaging concepts, link the relevant parts to the clinical research context in BRIDG, and thereby allow users to identify which imaging studies to pursue at a more detailed level in DICOM.