Restructuring Clinical Research Data Using Referent Tracking


Werner Ceusters

Center of Excellence in Bioinformatics and Life Sciences and Institute for Healthcare Informatics,
University at Buffalo, NY, USA




Combining clinical research data collections that cover the same domain but are compiled independently from each other, has the potential for providing answers with sufficient reliability and statistical significance to clinical questions that otherwise would remain unanswered. This tutorial, geared towards a wide audience with interest in ontology-based data mapping, will address the sort of problems that may be encountered in such endeavor, as well as the methodologies that can be used to solve them. Specific topics covered are problem areas such as ambiguities in data repositories, shortcomings of terminologies and inadequate semantics of representation languages. Attendees will learn how to overcome these problems by using the principles of Ontological Realism and Referent Tracking, and by becoming familiar with the formulation of bridging axioms between data sets and ontologies. The methodology will be explained using examples from the OPMQoL project, which involves data from five countries and 2000 patients, and illustrates how data from similar domain can be integrated to provide unambiguous meaning to instance data.


Outline of topics

  • Ambiguities in data repositories

  • Principles of Ontological Realism

  • Building an ontology for integrating clinical datasets

  • Principles of Referent Tracking

  • Data sets/ontology bridging


Learning objectives

  1. understand better the added value of the Ontological Realism principles over mere computational and logical frameworks,

  2. apply the principles to build or evaluate ontologies,

  3. assess how to optimally use such ontologies in eHealth Technologies such as EHRs, Dental Record systems, clinical research systems, and data warehouses, and

  4. make recommendations to clinicians and biomedical informaticists to improve the systems they are working with.


Target audience

The tutorial is suited for a wide range of participants with diverse backgrounds. Clinical researchers will acquire knowledge in how to compile and document de novo data collections that can be more reliably compared with similar resources, or learn how to make existing collections more suitable for that goal. Developers of semantic technologies, including terminologies and ontologies, will obtain valuable insight in how to apply the principles underlying Ontological Realism as a methodology to avoid mistakes that cannot be detected by logical formalisms alone. Users of eHealth applications in general, and clinicians in particular, specifically if they wish to function as champions ('super users') in Electronic Health or Dental Records or data warehouse customization efforts will learn the requirements such technologies must adhere to in order to make optimal use of ontologism.

Prerequisite knowledge

This tutorial requires of the attendees a thorough background in either any of the healthcare professions (medicine, nursing, dentistry) or informatics (computer science). Additional familiarity with healthcare informatics is advisable. No background in semantic technologies or ontology development is however necessary. It will be easier to follow the tutorial if attendees have read the references.