- Research Article
- Open Access
Semantic enrichment of longitudinal clinical study data using the CDISC standards and the semantic statistics vocabularies
© Leroux and Lefort; licensee BioMed Central. 2015
- Received: 8 August 2014
- Accepted: 5 March 2015
- Published: 9 April 2015
There is an increasing recognition of the need for the data capture phase of clinical studies to be improved and for more effective sharing of clinical data. The Health Care and Life Sciences community has embraced semantic technologies to facilitate the integration of health data from electronic health records, clinical studies and pharmaceutical research. This paper explores the integration of clinical study data exchange standards and semantic statistic vocabularies to deliver clinical data as linked data in a format that is easier to enrich with links to complementary data sources and consume by a broad user base.
We propose a Linked Clinical Data Cube (LCDC), which combines the strength of the RDF Data Cube and DDI-RDF vocabulary to enrich clinical data based on the CDISC standards. The CDISC standards provide the mechanisms for the data to be standardised, made more accessible and accountable whereas the RDF Data Cube and DDI-RDF vocabularies provide novel approaches to managing large volumes of heterogeneous linked data resources.
We validate our approach using a large-scale longitudinal clinical study into neurodegenerative diseases. This dataset, comprising more than 1600 variables clustered in 25 different sub-domains, has been fully converted into RDF forming one main data cube and one specialised cube for each sub-domain. One sub-domain, the Medications specialised cube, has been linked to relevant external vocabularies, such as the Australian Medicines Terminology and the ATC DDD taxonomy and DrugBank terminology. This provides new dimensions on which to query the data that promote the exploration of drug-drug and drug-disease interactions.
This implementation highlights the effectiveness of the association of the semantic statistics vocabularies for the publication of large heterogeneous data sets as linked data and the integration of the semantic statistics vocabularies with the CDISC standards. In particular, it demonstrates the potential of the two vocabularies in overcoming the monolithic nature of the underlying model and improving the navigation and querying of the data from multiple angles to support richer data analysis of clinical study data. The forecasted benefits are more efficient use of clinicians’ time and the potential to facilitate cross-study analysis.
- Semantic enrichment
- Longitudinal clinical study
- RDF data cube
- Medication mapping
In the last decade, the Health Care and Life Sciences community and pharmaceutical industry have wholeheartedly adopted  clinical study data exchange technologies based on XML to capture clinical study data. This is largely due to the recent strategy  of the Food and Drug Administration (FDA) in promoting the Clinical Data Interchange Standards Consortium (CDISC) suite of standards to facilitate data submission and exchange. Furthermore, the move by EU and US regulating bodies to open access to clinical data [3,4] will also foster the adoption of tools supporting clinical data management standards, especially those that can easily be linked to methods and tools developed for Government Linked Data and Linked Science Data.
CDISC has developed a set of platform-independent data standards  for the collection and dissemination of clinical trial data. The CDISC Operational Data Model (ODM) is an XML format that facilitates the exchange of clinical data captured during a clinical study. ODM-based files contain the study data and the associated descriptions of the data items, their groupings into Case Report Forms (CRFs), which are electronic documents to record the study data, and the associated questions and code lists. Furthermore, the FDA has mandated the use of other CDISC standards in clinical studies. In particular the CDISC Study Data Tabulation Model (SDTM) is used to facilitate study metadata submissions and improve the accountability of the study data. The role of the CDISC Clinical Data Acquisition Standards Harmonization (CDASH) is to standardise the generation of CRFs for clinical studies. The implementation of the ODM, STDM and CDASH standards in Clinical Data Management Systems (CDMS) has enabled larger and more diverse longitudinal clinical research studies and increased the capability of users to exchange and combine data .
Challenges relating to the cross-study analysis of clinical study data
A number of limitations relating to the reporting of results derived from current clinical trial endeavours were identified by van Valkenhoef et al. . In particular, they stress that: “current infrastructure is focused on text-based reports of single studies, whereas efficient evidence-based medicine requires the automated integration of multiple clinical trials from different information resources” . They specifically advocate for a comprehensive record of clinical trials to be made available in a machine understandable format that would improve the efficiency of evidence-based decision making but more importantly that decisions could then finally be explicitly linked back to the underlying data. Chief among their list of topics for future research directions are: (i) the development of a comprehensive data model for clinical trials and their aggregate level results; and (ii) the development of a platform to share structured systematic review data sets.
Our contribution: semantic enrichment
This research builds upon existing work  to semantically enrich longitudinal clinical study data, based on the CDISC standards, using semantic statistic vocabularies, namely the RDF Data Cube and DDI-RDF vocabularies. We propose a Linked Clinical Data Cube, a set of modular data cubes that helps manage the multi-dimensional and multi-disciplinary nature of the clinical data. The RDF Data Cube vocabulary  is used to build multi-dimensional data cubes and supports flexible access to the data via thematic slices. The DDI-RDF Discovery vocabulary  is effective at encoding the study-specific data dictionary embedded in the CDISC ODM standard as linked data and helps in managing the link between the data cube variables and the data.
Our objective is to make the data captured within the Australian, Imaging, Biomarker and Lifestyle study of Ageing (AIBL)  seamlessly available to researchers who wish to engage in cross-domain analysis of the data. We achieve our goal by semantically enriching the data, when possible, with external data sources. Our approach is four-fold: Phase 1: Integrating the CDISC ODM data model with the semantic statistic vocabularies. We describe how the clinical data available in CDISC ODM can be mapped to the RDF Data Cube and DDI-RDF Discovery vocabulary to form the Linked Clinical Data Cube. Phase 2: Splitting the data into modularised cubes. We outline the design principles of splitting the data into more modularised and manageable groupings to provide alternative mechanisms for accessing and querying the data. The RDF Data Cube and DDI-RDF vocabularies are pivotal elements of our slicing strategy and of the URI scheme defined for our implementation. Phase 3: Enriching the LCDC with the CDISC standards. We discuss how useful the benefits of clinical study data to adhering to the CDISC CDASH and SDTM standards then elaborate on guidelines to classify the data into the broad categories. Phase 4: Mapping the data to drug terminologies. We demonstrate the utility of the LCDC by mapping the medications data derived from the AIBL study to selected online drug terminologies.
The AIBL study
Number of instances for the LCDC classes organised by theme
In the remainder of this article, we outline an approach to semantically enrich clinical study data, in particular patient-reported medication usage, and facilitate their delivery to clinical researchers. In particular, we outline how the use of semantic statistics vocabularies is effective at organising the data into a LCDC. We also elaborate on the approach taken to categorise the AIBL data set into CDISC CDASH and SDTM domains and the work carried out to translate the CDISC standards into RDF. This leads into the discussion on the design principles for the LCDC and of the benefits of splitting the data into more modularised groupings.
The LCDC  comprises one main cube and several specialised cubes, one for each domain within the study, that integrates the CDISC ODM data model with the RDF Data Cube and DDI-RDF vocabularies. We elaborate further on the rationale behind this integration below. The LCDC is designed around a set of cubes, slices, observation groups and observations and these are discussed further below. The ability to standardise the clinical data in order to facilitate cross-domain and, possibly, cross-study analysis of the data is one of the salient objectives of the LCDC. To this end, we describe how the study variables have been enriched by the CDISC CDASH and SDTM standards. Aside from providing a standardised representation to the study variables and grouping them along the various CDISC categories, this enrichment process allows for seamless substitution of variable names in the navigation and querying of the clinical study. Finally, we outline how the coupling of the study data with external resources - in this case drug terminologies - can be achieved within the LCDC and we elaborate on our process to implement a linked medications data set and how the patient-reported medication intake from the AIBL study has been mapped to this data set.
Phase 1: Integrating the CDISC ODM data model with the semantic statistic vocabularies
Clinical study data is extracted in CDISC ODM format. The primary dimensions of the CDISC ODM data model are the Subject and Study Event of interest within the study. The additional dimensions, including the Study, Form, ItemGroup and Item, depend on the study domains and are specified by the data dictionary that defines the study. The strength of the RDF Data Cube is that the original structure of the CDISC ODM data model (Study-Subject-StudyEvent-Form-ItemGroup-Item) lends itself to be replicated in the generated cube with relative ease. A further contribution of the RDF Data Cube is that it can help overcome the monolithic nature of the ODM data model by facilitating the construction of multi-dimensional cubes that offer access points to the data via thematic slices. The LCDC is organised into one main cube and several specialised cubes corresponding to the various domains in the study.
Phase 2: Splitting the data into modularised cubes
The design of the LCDC is achieved in three steps. The first step involves splitting the dataset into smaller, more manageable specialised cubes. The second step is to define several slice hierarchies that offer multiple access options to the individual data records. The third step is to define a URI scheme that supports access to the cube at all levels of the slice hierarchy. These three steps are discussed below.
The LCDC defines three categories of slices. The time-series slices address the longitudinal nature of the study and organise the data into time-intervals and dated and non-dated time points. Cross-section slices adopt a subject-centric approach to the abstraction of the data along some important concepts such as gender, genotype and neurological classifications. The Theme slices categorise the data into the study domains and sub-domains (disco:Universe in DDI-RDF) and help link the main and specialised cubes together. This process enhances the navigation and querying of the data in the LCDC because we provide three direct links to nodes containing the data instead of one through the Phase series (at the level of the Study Event data in ODM), the Subject section (at the Subject level) and the sub-theme slice (at the Item Group level).
The URI scheme describing the LCDC follows the convention from the Linked Data API , which uses URIs ending with an identifier to provide access to a single instance (Item endpoint) and URIs ending with a keyword to provide access to a list of instances (List endpoint).
Phase 3: Enriching the LCDC with the CDISC standards
The CDISC CDASH and SDTM standards provide the means to standardise the clinical data. Despite not being designed around the CDISC standards, there is a good overlap between the AIBL study and the CDISC CDASH and SDTM standards for categories such as Vital Signs, Blood (represented by Laboratory Test in CDASH) and Medical History. For some categories within AIBL, the study data is clustered across many classes that do not necessarily fit to single CDASH or SDTM categories. We have chosen to map our medication data to the Concomitant Medications (CM) class within CDASH. Regarding CM, the approach taken by CDISC is to provide a framework and allow the users the ability to define the terminology of their choice. The AIBL Demographics data can be mapped to the CDISC Demographics and Subject Characteristics categories. SDTM’s Trial Arms, Trial Summary, Trial Visits and Subject Visits categories are appropriate targets for mapping longitudinal aspects of the study. For data items that are based on questionnaires, the methodology adopted by CDISC is to guide the user by providing a Questionnaire Supplements (QS) template that the user can mould to their needs. The SDTM standard provides approximately 50 questionnaires within the QS model that the user can use to model their study. The relatively low number of publicly available questionnaires is due to the fact that many of the questionnaires in clinical studies are licensed.
We have coupled the AIBL-specific variables to existing CDISC concepts, when possible, to allow a straightforward swap of variable names in a query. For example, the AIBL property for systolic blood Pressure (aiblvitalsigns:systolicBP) has been linked to the CDISC Vital Sign concept (cdiscvs:systolicBloodPressure).
Phase 4: Mapping the data to drug terminologies
In addition to the direct coupling between AIBL and CDISC definitions described above, we have mapped the patient-reported medication intake of the AIBL participants to three external terminologies: AMT, ATC DDDa and DrugBank. Our goal is to provide multiple links to hierarchical classifications of drugs. AMT provides unique codes and accurate standardised names to unambiguously identify all commonly used medicines in Australia with eight key top-level concepts . We augment AMT’s capabilities with links to ARTGb and UNIIc. ARTG contains the most comprehensive list of brand names (Trade Product) in Australia, while UNII provides a non-proprietary, unambiguous and unique list of substances as maintained by the FDA. DrugBank provides a rich taxonomy of drug information alongside comprehensive drug, gene and food interactions. The appeal for our project is in the exploration of drug-drug interactions that provide some insight into the potential risks and contraindications associated with the intake of the medication. Furthermore, by exploiting the gene-drug interactions of medication targets, we can extend our framework to support the discovery of biomarkers. Finally, the availability of the food interactions will be useful when we explore the association between the participant’s drug intake and type and amount of food consumed. Both ATC DDD and DrugBank provide a supplementary means to query the data. The five-level ATC DDD taxonomy of medications provides an additional mechanism for the data to be categorised and offers the means to aggregate the study data for statistical purposes. This is complementary to what is possible with the help of the vocabularies provided by AMT.
Medication mapping is challenging due to the quality, accuracy and completeness of the information. Previous studies [8,18] have identified numerous inconsistencies linked to the naming of the medications with a mix of trade name, active ingredients and informal name used to describe the prescribed medications.
Medications mapping statistics
Medicinal product/trade product/substance
The Linked Clinical Data Cube has been evaluated using the full AIBL data set to demonstrate its potential in formulating queries across the broad spectrum of tests and the categories within the clinical study. While simple queries can be answered using a single data cube, more complex queries need data from several cubes to be available. The clinical data is formalised into RDF prior to being loaded in a Virtuoso triple-store.
To demonstrate the utility of the LCDC, we have devised a set of three questions that are typical of the questions that the AIBL researchers are likely to ask of the study data. We provide, below, a listing of the three queries. However, due to privacy constraints, we have structured our queries so that they only return aggregated counts because we are unable to present the participants unique identifier as part of the results of the queries.
Those SPARQL queries have been chosen in order to demonstrate the breadth and depth of questions that may be asked on the data set. They demonstrate how data from the AIBL study can be effortlessly combined with drug information, for example, in order to facilitate queries that answer questions based on drug classifications. Furthermore, we also demonstrate, through the integration of the AIBL data set with terminologies from the CDISC standards, how the AIBL data set can be queried by using the CDISC standardised terminology rather than the actual test names used by the AIBL study. We believe that these types of queries will drive the cross-study and cross-domain benefits of the linked clinical data approaches such as the LCDC.
Query 1: Using CDISC terms, find the number of participants who have hypertension
The query obtains the relevant test names from the ontology by performing a lookup of properties that are sub-properties of the CDISC Vital Signs (prefix: vs) diastolic and systolic blood pressure variables. This is achieved by this statement:
This query is possible because we have implemented a linked set that connects the variable name from the AIBL study to the standardised terminology in CDISC SDTM vs domain as illustrated below.
We believe that the use of linksets in this manner is important and useful because it adheres to the principles of information hiding in that the user need not be aware of the exact wording of a variable. As long as the user knows the corresponding standardised variable name, the user is able to successfully execute a query on the data set. We intend to further develop this traceability mechanism with the help of the Provenance Ontology  to fully disclose how the published data is derived from the originally captured data.
The result of Query 1 is displayed below:
Query 2: How many participants are taking an anti-diabetic drug such as Metformin?
Some studies [21,22] have shown a possible link between type2 diabetes and early-stage AD. In this query, we retrieve a list of anti-diabetic drugs to demonstrate the benefits of linking the patient-reported medications to standardised external terminologies and the strength of the LCDC in using federated queries to facilitate cross-domain querying. The first portion of this query obtains a list of anti-diabetic drugs from DrugBank (outlined in section A in the SPARQL). The second part of the query utilises the mappings between the patient-reported medications and DrugBank entities to link to the anti-diabetic drugs identified in section A.
The significance of this mapping is the provision of drug-drug, drug-gene and possibly drug-disease and gene-gene information relating to the AIBL study to the researchers by fully utilising the links provided by DrugBank.
Participants taking anti-diabetic drugs
Query 3: Are there participants whose classification has transitioned from healthy to mild cognitive impairment but whose triglyceride’s level has remained normal?
Research has investigated the risk factors associated with low-density lipoproteins or triglycerides on the incidence and progression of dementia and AD in later life . With this in mind, we construct the query below to retrieve participants’ records whose confirmed classification status have been updated from being healthy as subjective memory complainer or non-memory complainer to having mild cognitive impairment but who have also maintained a normal (< 1.7 mmol/L) level of triglycerides in their blood sample over the course of an 18-month period between the baseline and 18-month time-points.
Participants’ classifications and triglycerides level
Execution time (msec)
This paper has outlined the semantic enrichment of longitudinal clinical study data based on the CDISC standards with elements from the semantic statistics vocabularies, namely the RDF Data Cube and the DDI-RDF Discovery vocabularies. We have outlined how the Health Care and Life Science community is likely to benefit from the adoption of tools and techniques that will deliver clinical data as linked data and advance its integration with complementary data sources. In this regard, we have proposed a Linked Clinical Data Cube, which integrates one main and several specialised data cubes to provide increased flexibility in the navigation of the clinical data and allow the users to formulate the queries more efficiently and effectively. The Linked Clinical Data Cube combines the strength of the RDF Data Cube in defining multi-dimensional data cubes and the DDI-RDF Discovery vocabulary in encoding the CDISC metadata and the study specific data dictionary as linked data. Our approach was validated using data captured as part of a longitudinal clinical study into neurodegenerative diseases. This research has resulted in four contributions. First, we have uncovered the complementarities of the RDF Data Cube and DDI-RDF Discovery vocabularies for the publication of large heterogeneous data sets as linked data. Second, we have demonstrated the fit of the semantic statistics vocabularies to enrich the CDISC ODM data model for the publication of clinical study data as linked data. Third, we have illustrated how the clinical study data has been semantically enriched with links to external resources and how they ultimately improve the navigation and querying of the data. Fourth, we have built the foundations of a framework supporting cross-domains and cross-study analysis by adopting a more standardised data structure. Our next step is to enrich the remaining study data set with concepts from other domain ontologies, such as Blood, Neuropsychological tests and Nutrition, to name just three.
a Anatomical Therapeutic Chemical Defined Daily Dose.
b Australian Register of Therapeutic Goods.
c Unique Ingredient Identifier.
d 30560011000036108 | trade product |.
e 30497011000036103 | medicinal product|.
f 30388011000036105 | medicinal substance |.
g Systematized Nomenclature of Medicine Clinical Terms.
The authors would like to express their gratitude to Drs Alejandro Metke and Michael Lawley for their assistance in scoping the Medications case study and along with Dr Bevan Koopman and Mr Simon McBride for reviewing the paper and to Mr Simon Gibson and Mr Louis Delachat for their assistance in the project.
- Lu Z, Su J. Clinical data management: Current status, challenges, and future directions from industry perspectives. Open Access J Clin Trials. 2010; 2:93–105.View ArticleGoogle Scholar
- FDA: Data Standards Strategy Version 1.0. 2012. http://www.fda.gov/downloads/Drugs/DevelopmentApprovalProcess/FormsSubmissionRequirements/ElectronicSubmissions/UCM332955.pdf.
- Eichler H-G, Abadie E, Breckenridge A, Leufkens H, Rasi G. Open clinical trial data for all? A view from regulators. PLoS Med. 2012; 9(4):1001202.View ArticleGoogle Scholar
- IOM. Sharing Clinical Research Data - Workshop Summary. The National Academies Press. 2013.:9–17. http://www.iom.edu/Reports/2013/ Sharing-Clinical-Research-Data.aspx Sharing-Clinical-Research-Data.aspx.
- van Valkenhoef G, Tervonen T, de Brock B, Hillege H. Clinical trials information in drug development and regulation: existing systems and standards. Netherlands: University of Groningen Press; 2012.Google Scholar
- Ohmann C, Kuchinke W. Future developments of medical informatics from the viewpoint of networked clinical research. Methods Inf Med. 2009; 48(1):45–54.Google Scholar
- van Valkenhoef G, Tervonen T, de Brock B, Hillege H. Deficiencies in the transfer and availability of clinical trials evidence: A review of existing systems and standards. BMC Med Informatics Decision Making. 2012; 12(1):95.View ArticleGoogle Scholar
- Leroux H, Lefort L. Using CDISC ODM and the RDF Data Cube for the semantic enrichment of longitudinal clinical trial data. In: Proceedings of the 5th International Workshop on Semantic Web Applications and Tools for Life Sciences. Paris, France: CEUR Workshop Proceedings: 2012.Google Scholar
- Cyganiak R, Reynolds D, Tennison J. The RDF data cube vocabulary. World Wide Web Consortium. 2014. http://www.w3.org/TR/vocab-data-cube/.
- Bosch T, Cyganiak R, Zapilko B, Cotton F, Gregory A, Kämpgen B, et al.DDI-RDF Discovery vocabulary a vocabulary for publishing metadata about data sets (research and survey data) into the web of linked data unofficial draft. Technical report, DDI Alliance; June 2013.Google Scholar
- Ellis KA, Bush AI, Darby D, De Fazio D, Foster J, Hudson P, et al.The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging: methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease. Int Psychogeriatrics/IPA. 2009; 21(4):672–87.View ArticleGoogle Scholar
- Ellis KA, Rainey-Smith SR, Rembach A, Macaulay SL, Villemagne VL. Enabling a multidisciplinary approach to the study of ageing and alzheimer’s disease: An update from the Australian Imaging Biomarkers and Lifestyle (aibl) study. Int Rev Psychiatry. 2013; 25(6):699–710.View ArticleGoogle Scholar
- Leroux H, McBride S, Gibson S. On selecting a clinical trial management system for large scale, multi-centre, multi-modal clinical research study. Stud Health Technol Informatics. 2011; 168:89–95.Google Scholar
- OpenClinica https://www.openclinica.com/.
- Lefort L, Leroux H. Design and generation of Linked Clinical Data Cubes. In: Proceedings of 1st International Workshop on Semantic Statistics (SemStats 2013). Sydney, Australia: 2013.Google Scholar
- Linked Data API https://code.google.com/p/linked-data-api/.
- Michel J, Lawley MJ, Chu A, Barned J. Mapping the queensland health ipharmacy medication file to the Australian Medicines Terminology using Snapper. Stud Health Technol Informatics. 2011; 168:104–16.Google Scholar
- McBride SJ, Lawley MJ, Leroux H, Gibson S. Using Australian Medicines Terminology (AMT) and SNOMED CT-AU to better support clinical research. Stud Health Technol Informatics. 2012; 178:144–9.Google Scholar
- High Blood Pressure http://www.nlm.nih.gov/medlineplus/highbloodpressure.html.
- PROV-O: The PROV Ontology http://www.w3.org/TR/prov-o/.
- Griggs J. Eating your way to dementia. New Scientist. 2013; 220(2945):6–7.View ArticleGoogle Scholar
- Kloppenborg RP, van den Berg E, Kappelle LJ, Biessels GJ. Diabetes and other vascular risk factors for dementia: which factor matters most? A systematic review. Eur J Pharmacol. 2008; 585(1):97–108.View ArticleGoogle Scholar
- Willighagen EL, Waagmeester A, Spjuth O, Ansell P, Williams AJ, Tkachenko V, et al.The CHEMBL database as linked open data. J Cheminformatics. 2013; 5(1):1–12.View ArticleGoogle Scholar
- Groth P, Loizou A, Gray AJ, Goble C, Harland L, Pettifer S. API-centric linked data integration: The open PHACTS discovery platform case study. Web Semantics: Science, Services and Agents on the World Wide Web. 2014; 29(0):12–8. http://dx.doi.org/10.1016/j.websem.2014.03.003, http://www.sciencedirect.com/science/article/pii/S1570826814000195.View ArticleGoogle Scholar
- Dumontier M, Villanueva-Rosales N. Towards pharmacogenomics knowledge discovery with the semantic web. Brief Bioinform. 2009; 10(2):153–63.View ArticleGoogle Scholar
- Marshall MS, Boyce R, Deus HF, Zhao J, Willighagen EL, Samwald M, et al.Emerging practices for mapping and linking life sciences data using rdf - a case series. Web Semantics: Sci, Services Agents World Wide Web. 2012; 14(0):2–13. Special Issue on Dealing with the Messiness of the Web of Data.View ArticleGoogle Scholar
- Luciano J, Andersson B, Batchelor C, Bodenreider O, Clark T, Denney C, et al.The translational medicine ontology and knowledge base: driving personalized medicine by bridging the gap between bench and bedside. J Biomed Semantics. 2011; 2(Suppl 2).Google Scholar
- Abler D, Crichton C, Welch J, Davies J, Harris S. Models for forms. In: Proceedings of the Compilation of the Co-located Workshops on DSM’11, TMC’11, AGERE!’11, AOOPES’11, NEAT’11, & VMIL’11. NY, USA: ACM New York: 2011. p. 13–8.Google Scholar
- Bioportal Ontologies http://bioportal.bioontology.org/.
- Sim I, Carini S, Tu S, Wynden R, Pollock BH, Mollah SA, et al.The human studies database project: federating human studies design data using the ontology of clinical research. AMIA Summits Transl Sci Proc. 2010; 2010:51–5.Google Scholar
- Ogbuji C. A framework ontology for computer-based patient record systems. In: Proceedings of the ICBO: International Conference on Biomedical Ontology: 2011. p. 217–23.Google Scholar
- Megan Kong Y, Dahlke C, Xiang Q, Qian Y, Karp D, Scheuermann RH. Toward an ontology-based framework for clinical research databases. J Biomed Informatics. 2011; 44(1):48–58.View ArticleGoogle Scholar
- Samwald M, Jentzsch A, Bouton C, Stie Kallesøe C, Willighagen E, Hajagos J, et al.Linked open drug data for pharmaceutical research and development. J Cheminformatics. 2011; 3(1):19.View ArticleGoogle Scholar
- Williams AJ, Harland L, Groth P, Pettifer S, Chichester C, Willighagen EL, et al.Open PHACTS: semantic interoperability for drug discovery. Drug Discovery Today. 2012; 17(21):1188–98.View ArticleGoogle Scholar
- Belleau F, Nolin M-A, Tourigny N, Rigault P, Morissette J. Bio2RDF: Towards a mashup to build bioinformatics knowledge systems. J Biomed Informatics. 2008; 41(5):706–16.View ArticleGoogle Scholar
- Pence HE, Williams A. Chemspider: An online chemical information resource. J Chem Educ. 2010; 87(11):1123–4.View ArticleGoogle Scholar
- Willighagen EL, Alvarsson J, Andersson A, Eklund M, Lampa S, Lapinsh M, et al.Linking the resource description framework to cheminformatics and proteochemometrics. J Biomed Semantics. 2011; 2(S-1):6.View ArticleGoogle Scholar
- Bilder G. Content negotiation for CrossRef DOIs. 2011. http://crosstech.crossref.org/2011/04/content_negotiation_for_crossr.html.
- Laleci GB, Yuksel M, Dogac A. Providing semantic interoperability between clinical care and clinical research domains. Biomed Health Informatics, IEEE J. 2013; 17(2):356–369.View ArticleGoogle Scholar
- Neumayr B, Schrefl M, Linner K. Semantic Cockpit: An Ontology-Driven, Interactive Business Intelligence Tool for Comparative Data Analysis In: De Troyer O, Bauzer Medeiros C, Billen R, Hallot P, Simitsis A, Van Mingroot H, editors. Advances in Conceptual Modeling. Recent Developments and New Directions SE - 9. Lecture Notes in Computer Science. Springer Berlin Heidelberg: Springer: 2011. p. 55–64.Google Scholar
- Kamateri E, Kalampokis E, Tambouris E, Tarabanis K. The linked medical data access control framework. J Biomed Informatics. 2014; 50(0):213–225.View ArticleGoogle Scholar
- PhUSE Wiki. http://www.phusewiki.org/wiki/index.php?title=Semantic_Technology.
- FDA/PhUSE: Representing Existing CDISC Standards in RDF. https://github.com/phuse-org/rdf.cdisc.org.
- NCI: CDISC Terminology. http://www.cancer.gov/cancertopics/cancerlibrary/terminologyresources/cdisc.
- Andersen M. Linked data to support clinical and non-clinical reporting. In: Proceedings of 2nd International Workshop on Semantic Statistics (SemStats 2014). Riva del Garda, Italy: 2014.Google Scholar
- Williams T. A primer on converting analysis results data to RDF Data Cubes using free and open source tools. In: Proceedings of 10th Annual PhUSE conference(PhUSE 2014). London, UK: 2014.Google Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.