Synonyms
Clinical content database; Clinical content registry; Clinical content repository; Clinical knowledge base; Clinical knowledge directory; Clinical knowledge management repository
Definition
A clinical knowledge repository (CKR) is a multipurpose storehouse for clinical knowledge assets. “Clinical knowledge asset” is a generic term that describes any type of human or machine-readable electronic content used for computerized clinical decision support. A CKR is normally implemented as an enterprise resource that centralizes a large quantity and wide variety of clinical knowledge assets. A CKR provides integrated support to all asset lifecycle phases such as authoring, review, activation, revision, and eventual inactivation. A CKR routinely provides services to search, retrieve, transform, merge, upload, and download clinical knowledge assets. From a content curation perspective, a CKR has to ensure proper asset provenance, integrity, and versioning, along with effective access and utilization constraints compatible with collaborative development and deployment activities. A CKR can be considered a specialized content management system, designed specifically to support clinical information systems. Within the context of clinical decision support systems, a CKR can be considered a special kind of knowledge base designed to manage multiple types of human and machine-readable clinical knowledge assets.
Key Points
In recent years, multiple initiatives have attempted to better organize, filter, and apply the ever-growing biomedical knowledge. Among these initiatives, one of the most promising is the utilization of computerized clinical decision support systems. Computerized clinical decision support can be defined as computer systems that provide the correct amount of relevant knowledge at the appropriate time and context, contributing to improved clinical care and outcomes. A wide variety of knowledge-driven tools and methods have resulted in multiple modalities of clinical decision support, including information selection and retrieval, information aggregation and presentation, data entry assistance, event monitors, care workflow assistance, and descriptive or predictive modeling. A CKR provides an integrated storage platform that enables the creation and maintenance of multiple types of knowledge assets. A CKR ensures that different modalities of decision support can be combined to properly support the activities of clinical workers. Core requirements guiding the implementation of a CKR include clinical knowledge asset provenance (metadata), versioning, and integrity. Other essential requirements include the proper representation of access and utilization constraints, taking into account the collaborative nature of asset development processes and deployment environments. Another fundamental requirement is to aptly represent multiple types of knowledge assets, where each type might require specialized storage and handling. The CKR core requirements are similar to those specified for other types of repositories used for storage and management of machine-readable assets.
Historical Background
Biomedical knowledge has always been in constant expansion, but unprecedented growth is being observed during the last decade. Over 32% of the 22.3 million citations accumulated by MEDLINE until January of 2014 were created in the last 10 years, with an average of over 708,400 new citations per year [1]. The number of articles published each year is commonly used as an indicator of how much new knowledge the scientific community is creating. However, from a clinical perspective, particularly for those involved with direct patient care, the vast amount of new knowledge represents an ever-growing gap between what is known and what is routinely practiced. Multiple initiatives in recent years have attempted to better organize, filter, and apply the knowledge being generated. Among these various initiatives, one of the most promising is the utilization of computerized clinical decision support systems [2]. In fact, some authors avow that clinical care currently mandates a degree of individualization that is inconceivable without computerized decision support [3].
Computerized clinical decision support can be defined as computer systems that provide the correct amount of relevant knowledge at the appropriate time and context, ultimately contributing to improved clinical care and outcomes [4]. Computerized clinical decision support has been an active area of informatics research and development for the last three decades [5]. A wide variety of knowledge-driven tools and methods have resulted in multiple modalities of clinical decision support, including information selection and retrieval (e.g., infobuttons, crawlers), information aggregation and presentation (e.g., summaries, reports, dashboards), data entry assistance (e.g., forcing functions, calculations, evidence-based templates for ordering and documentation), event monitors (e.g., alerts, reminders, alarms), care workflow assistance (e.g., protocols, care pathways, practice guidelines), and descriptive or predictive modeling (e.g., diagnosis, prognosis, treatment planning, treatment outcomes). Each modality requires specific types of knowledge assets, ranging from production rules to mathematical formulas, and from automated workflows to machine learning models. A CKR provides an integrated storage platform that enables the creation and maintenance of multiple types of assets using knowledge management best practices [6].
The systematic application of knowledge management processes and best practices to the biomedical domain is a relatively recent endeavor [5]. Consequently, a CKR should be seen as an evolving concept that is progressively being recognized as a fundamental component for the acquisition, storage, and maintenance of clinical knowledge assets. Most clinical decision support systems currently in use still rely on traditional knowledge bases that handle a single type of knowledge asset and do not provide direct support for a complete asset lifecycle. Another important principle is the recognition that different modalities of decision support have to be combined and subsequently integrated with information systems to properly support the activities of clinical workers. The premise of integrating multiple modalities of clinical decision support reinforces the need for knowledge management processes supported by a CKR.
Foundations
Core requirements guiding the implementation of a CKR include clinical knowledge asset provenance (metadata), versioning, and integrity. Requirements associated with proper access and utilization constraints are also essential, particularly considering the collaborative nature of most asset development processes and deployment environments. Another fundamental requirement is to aptly represent multiple types of knowledge assets, where each type might require specialized storage and handling. The CKR core requirements are generally similar to those specified for other types of repositories used for storage and management of diverse machine-readable assets.
Requirements associated with asset provenance can be implemented using a rich set of metadata properties that describe the origin, purpose, evolution, and status of each clinical knowledge asset. The metadata properties should reflect the information that needs to be captured during each phase of the knowledge asset lifecycle process, taking into account multiple iterative authoring and review cycles, followed by a possibly long period of clinical use that might require multiple periodic revisions (updates). Despite the diversity of asset types, each with a potentially distinct lifecycle process, a portion of the metadata properties should be consistently implemented, enabling basic searching and retrieval services across asset types. Ideally, the shared metadata should be based on metadata standards (e.g., “Dublin Core Metadata Element Set” (http://dublincore.org/documents/dces/)). The adoption of standard metadata properties also simplifies the integration of external collections of clinical knowledge assets in a CKR. In addition to a shared set of properties, a CKR should also accommodate extended sets of properties specific for each clinical knowledge asset type and its respective lifecycle process. Discrete namespaces are commonly used to represent type-specific extended metadata properties.
Asset version and status, along with detailed change tracking, are vital requirements for a CKR. Different versioning strategies can be used, but as a general rule there should be only one clinically active version of any given knowledge asset. This general rule is easily observed if the type and purpose of the clinical knowledge asset remains the same throughout its lifecycle. However, a competing goal is created with the very desirable evolution of human-readable assets to become machine-readable. Such evolution invariably requires the creation of new knowledge assets of different types and potentially narrower purposes. In order to support this “natural” evolution, a CKR should implement the concept of asset generations, while preserving the change history that links one generation to the next. Also within a clinical setting, it is not uncommon to have to ensure that knowledge assets comply with, or directly implement, different norms and regulations. As a result, the change history of a clinical knowledge asset should identify the standardization and compliance aspects considered, enabling subsequent auditing and/or eventual certification.
Ensuring the integrity of clinical knowledge assets is yet another vital requirement for a CKR. Proper integrity guarantees that each asset is unique within a specific type and purpose, and that all its required properties are accurately defined. Integrity requirements also take into account the definition and preservation of dependencies between clinical knowledge assets. These dependencies can be manifested as simple hyperlinks, or as integral content defined as another independent asset. Creating clinical knowledge assets from separate components or modules (i.e., modularity) is a very desirable feature in a CKR - one that ultimately contributes to the overall maintainability of the various asset collections. However, modularity introduces important integrity challenges, particularly when a new knowledge asset is being activated for clinical use. Activation for clinical use requires a close examination of all separate components, sometimes triggering unplanned revisions of components already in routine use. Another important integrity requirement is the ability to validate the structure and the content of a clinical knowledge asset against predefined templates (schemas) and dictionaries (ontologies). Asset content validation is essential for optimal integration with clinical information systems. Ideally, within a given healthcare organization all clinical information systems and the CKR should utilize the same reference ontologies.
Contextual characteristics of the care delivery process establish the requirements associated with proper access, utilization, and presentation of the clinical knowledge assets. The care delivery context is a multidimensional constraint that includes characteristics of the patient (e.g., gender, age group, language, clinical condition), the clinical worker (e.g., discipline, specialty, role), the clinical setting (e.g., inpatient, outpatient, ICU, Emergency Department), and the information system being used (e.g., order entry, documentation, monitoring), among others. The care delivery context normally applies to the entire clinical knowledge asset, directly influencing search, retrieval, and presentation services. The care delivery context can also be used to constrain specific portions of a knowledge asset, including links to other embedded assets, making them accessible only if the constraints are satisfied. An important integrity challenge created by the systematic use of the care delivery context is the need for reconciling conflicts caused by incompatible asset constraints, particularly when different teams maintain the assets being combined. In this scenario, competing requirements are frequently present, namely the intention to maximize modularity and reusability versus the need to maximize clinical specificity and ease or use.
The accurate selection, retrieval, and presentation of unstructured assets is generally perceived as a simple but very useful modality of clinical decision support, particularly if the information presented to the clinical worker is concise and appropriate to the care being delivered. However, the appropriateness of the information is largely defined by the constraints imposed by the aforementioned care delivery context. Moreover, the extent of indexing (“retrievability”) of most collections of unstructured clinical knowledge assets is not sufficient to fully recognize detailed care delivery context expressions. Ultimately, the care delivery context provides an extensible mechanism for defining the appropriateness of a given clinical knowledge asset in response to a wide variety of CKR service requests.
The requirements just described are totally or partially implemented as part of general-purpose (enterprise) content management systems. However, content management systems have been traditionally constructed for managing primarily human-readable electronic content. Human-readable content, more properly characterized as unstructured knowledge assets, include narrative text, diagrams, and multimedia objects. When combined, these unstructured assets likely represent the largest portion of the inventory of clinical knowledge assets of any healthcare institution. As a result, in recent years different healthcare organizations have deployed CKRs using enterprise content management systems, despite their inability to manage machine-readable content.
Key Applications
Computerized Clinical Decision Support, Clinical Knowledge Engineering, Clinical Information Systems.
Recommended Reading
Statistical Reports on MEDLINE®/PubMed® Baseline Data, National Library of Medicine, Department of Health and Human Services [Online]. Available at: http://www.nlm.nih.gov/bsd/licensee/baselinestats.html. Accessed 29 Jun 2014.
Wyatt JC. Decision support systems. J R Soc Med. 2000;93(12):629–33.
Bates DW, Gawande AA. Improving safety with information technology. N Engl J Med. 2003;348(25):2526–34.
Berner E. Clinical decision support systems: state of the art. Rockville: Agency for Healthcare Research and Quality; 2009. http://healthit.ahrq.gov/sites/default/files/docs/page/09-0069-EF_1.pdf. Accessed 29 Jun 2014.
Greenes RA, editor. clinical decision support: the road to broad adoption. 2nd ed. Burlington: Academic; 2014.
Rocha RA, Maviglia SM, Sordo M, Rocha BH. Clinical knowledge management program. In: Greenes RA, editor. Clinical decision support – the road to broad adoption. 2nd ed. Burlington: Academic; 2014. p. 773–817.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Section Editor information
Rights and permissions
Copyright information
© 2018 Springer Science+Business Media, LLC, part of Springer Nature
About this entry
Cite this entry
Rocha, R.A. (2018). Clinical Knowledge Repository. In: Liu, L., Özsu, M.T. (eds) Encyclopedia of Database Systems. Springer, New York, NY. https://doi.org/10.1007/978-1-4614-8265-9_57
Download citation
DOI: https://doi.org/10.1007/978-1-4614-8265-9_57
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-1-4614-8266-6
Online ISBN: 978-1-4614-8265-9
eBook Packages: Computer ScienceReference Module Computer Science and Engineering