Abstract:Introduction: This article is part of the Focus Theme of METHODS of Information in Medicine on "Managing Interoperability and Complexity in Health Systems". Background: The need for complementary access to multiple RDF databases has fostered new lines of research, but also entailed new challenges due to data representation disparities. While several approaches for RDF-based database integration have been proposed, those focused on schema alignment have become the most widely adopted. All state-of-the-art solutions for aligning RDF-based sources resort to a simple technique inherited from legacy relational database integration methods. This technique - known as element-to-element (e2e) mappings - is based on establishing 1:1 mappings between single primitive elements - e.g. concepts, attributes, relationships, etc. - belonging to the source and target schemas. However, due to the intrinsic nature of RDF - a representation language based on defining tuples < subject, predicate, object > -, one may find RDF elements whose semantics vary dramatically when combined into a view involving other RDF elements - i.e. they depend on their context. The latter cannot be adequately represented in the target schema by resorting to the traditional e2e approach. These approaches fail to properly address this issue without explicitly modifying the target ontology, thus lacking the required expressiveness for properly reflecting the intended semantics in the alignment information. Objectives: To enhance existing RDF schema alignment techniques by providing a mechanism to properly represent elements with context-dependent semantics, thus enabling users to perform more expressive alignments, including scenarios that cannot be adequately addressed by the existing approaches. Methods: Instead of establishing 1:1 correspondences between single primitive elements of the schemas, we propose adopting a view-based approach. The latter is targeted at establishing mapping relationships between RDF subgraphs - that can be regarded as the equivalent of views in traditional databases -, rather than between single schema elements. This approach enables users to represent scenarios defined by context-dependent RDF elements that cannot be properly represented when adopting the currently existing approaches. Results: We developed a software tool implementing our view-based strategy. Our tool is currently being used in the context of the European Commission funded p-medicine project, targeted at creating a technological framework to integrate clinical and genomic data to facilitate the development of personalized drugs and therapies for cancer, based on the genetic profile of the patient. We used our tool to integrate different RDF-based databases - including different repositories of clinical trials and DICOM images - using the Health Data Ontology Trunk (HDOT) ontology as the target schema. Conclusions: The importance of database integration methods and tools in the context of biomedical research has been widely recognized. Modern research in this area - e.g. identification of disease biomarkers, or design of personalized therapies - heavily relies on the availability of a technical framework to enable researchers to uniformly access disparate repositories. We present a method and a tool that implement a novel alignment method specifically designed to support and enhance the integration of RDF-based data sources at schema (metadata) level. This approach provides an increased level of expressiveness compared to other existing solutions, and allows solving heterogeneity scenarios that cannot be properly represented using other state-of-the-art techniques.

VIEW: a framework for organization level interactive record linkage to support reproducible data science

An Act Indexing Information Model for Clinical Data Integration

VBridge: Connecting the Dots Between Features and Data to Explain Healthcare Models.

Implementing a Hash-Based Privacy-Preserving Record Linkage Tool in the OneFlorida Clinical Research Network.

VIEWER: an extensible visual analytics framework for enhancing mental healthcare

HistoriView: Implementation and Evaluation of a Novel Approach to Review a Patient Using a Scalable Space-Efficient Timeline without Zoom Interactions

Manual evaluation of record linkage algorithm performance in four real-world datasets

LinkR: an open source, low-code and collaborative data science platform for healthcare data analysis and visualization

Interactive similar patient retrieval for visual summary of patient outcomes

Issues in identification and linkage of patient records across an integrated delivery system

CIDACS-RL: a novel indexing search and scoring-based record linkage system for huge datasets with high accuracy and scalability

Linkability measures to assess the data characteristics for record linkage

PNS215 IMPROVING ACCURACY OF RECORD LINKAGE USING GRAPH STRUCTURES: RELEVANCE FOR HEALTH OUTCOMES RESEARCH?

Toward a view-oriented approach for aligning RDF-based biomedical repositories

CollaborationViz: Interactive Visual Exploration of Biomedical Research Collaboration Networks

A Visual Analytic Tool (VIADS) to Assist the Hypothesis Generation Process in Clinical Research: Mixed Methods Usability Study

Privacy preserving interactive record linkage (PPIRL)

Distributed Record Linkage in Healthcare Data with Apache Spark

Efficient Techniques for Online Record Linkage

Bayesian Record Linkage with Variables in One File

Designing an Algorithm to Preserve Privacy for Medical Record Linkage with Error-Prone Data.