Publishing Chinese medicine knowledge as Linked Data on the Web

Abstract

Background: Chinese medicine (CM) draws growing attention from Western healthcare practitioners and patients. However, the integration of CM knowledge and Western medicine (WM) has been hindered by a barrier of languages and cultures as well as a lack of scientific evidence for CM's efficacy and safety. In addition, most of CM knowledge published with relational database technology makes the integration of databases even more challenging.

Methods: Linked Data approach was used in publishing CM knowledge. This approach was applied to publishing a CM linked dataset, namely RDF-TCM http://www.open-biomed.org.uk/rdf-tcm/ based on TCMGeneDIT, which provided association information about CM in English.

Results: The Linked Data approach made CM knowledge accessible through standards-compliant interfaces to facilitate the bridging of CM and WM. The open and programmatically-accessible RDF-TCM facilitated the creation of new data mash-up and novel federated query applications.

Conclusion: Publishing CM knowledge in Linked Data provides a point of departure for integration of CM databases.

Figure 1

The diagram of the RDF-TCM ontology. The diagram illustrates the main classes (the boxes) and object properties (the directed arrows) in the RDF-TCM ontology http://purl.org/net/tcm-onto/. The data properties of the ontology are not shown.

Figure 2

The data mash-up application for alternative medicines. A search for alternative medicines for the Alzheimer's disease takes a disease name as the input and search in the RDF-TCM dataset for a list of possible alternative medicine associated with the disease.

Figure 3

Detailed information about each alternative medicine. More information about Ginkgo biloba is returned, including its general information retrieved from DBpedia (left-side pane) and its putative effects information retrieved from RDF-TCM (right-side pane). This query demonstrates how we can create a more complete picture of knowledge about Ginkgo biloba by querying distributed linked datasets.

Figure 4

Clinical trials related to Ginkgo biloba. Clinical trials related to Ginkgo biloba are found from the LinkedCT dataset. These results are also linked to LinkedCT where more information about these trials can be found.

Figure 5

Confirmation of genetic evidences for the efficacy of alternative medicines using RDF-TCM and Diseasome. We first use the RDF-TCM dataset to find genes associated with the Alzheimer's diseases and the herb Ginkgo biloba, and we then use the Diseasome database to search for the diseases associated with these genes. If an RDF-TCM gene is also associated with the Alzheimer's disease according to Diseasome, we then confirm that gene as an Alzheimer's gene. In this way, we use two datasets created by two different medical research communities to confirm genetic evidence for the herbs.

Figure 6

Finding alternative medicines as well as their side effects powered by SQUIN. To find alternative medicines to Simvastatin as well as their side effects powered by SQUIN, we use a Linked Data query engine, which allows one SPARQL query to access 6 distributed linked datasets published at different sources, including Drugbank, Diseasome, SIDER, LinkedCT, Dailymed and RDF-TCM.