Efstathios Karanastasis

A novel semantic representation for eligibility criteria in clinical trials

Eligibility Criteria (EC) comprise an important part of a clinical study, being determinant of its cost, duration and overall success. Their formal, computer-processable description can significantly improve clinical trial design and conduction by enabling their intelligent processing, replicability and linkability with other data. For EC representation purposes, related standards were investigated, along with published literature. Moreover, a considerable number of clinicaltrials.gov studies was analyzed in collaboration with clinical experts for the determination and classification of parameters of clinical research importance. The outcome of this process was the EC Representation; a CDISC-compliant schema for organizing criteria along with a patient-centric model for their formal expression, properly linked with international classifications and codifications. Its evaluation against 200 randomly selected EC indicated that it can adequately serve its purpose, while it can be also combined with existing tools and components developed for both EC specification and especially application to Electronic Health Records.

Meaning Inference of Abbreviations Appearing in Clinical Studies

The number of publicly available clinical studies is constantly increasing, formulating a rather promising corpus of documents for clinical research purposes. However, the abbreviations used in these studies pose a serious barrier to any text mining technique. This paper presents a study conducted in the above domain, which used specifically developed tools and mechanisms in order to process a number of randomly selected documents from clinicaltrialsregister.eu. The analysis performed indicated that abbreviations appear at a large scale without their long form (aka expansion). In order to assess the abbreviations’ true meaning, it is necessary to utilize the appropriate corpus of documents, apply innovative algorithms and techniques to detect their possible expansions, and accordingly select the appropriate ones. Furthermore, the discrimination power of tokens has a distinctive role in abbreviations construction, and hence, it can facilitate the detection of acronym-type abbreviations. Additionally, the expressions in which abbreviations appear, as well as the preceding or following text are of primary importance for selecting the appropriate meaning.

The OpenScienceLink architecture for novel services exploiting open access data in the biomedical domain

The rapidly growing wealth of published scientific work, produced by researchers and scholars, has resulted in a pressing need for more effective processes towards reviewing scientific articles and research data, organizing data journals, as well as for improved tools and techniques for bibliographic analysis and management of scientometrics. The ongoing EU research project OpenScienceLink aims to provide a remedy for the aforementioned problems, as well as offer a wide range of opportunities for better collaboration between researchers, by introducing a web based Platform which offers efficient and intelligent added-value applications and services for exploiting open access scientific information in the biomedical domain. The Platform is empowered by the semantic and social networking capabilities of three leading edge background infrastructures, which have been adapted and integrated for the scopes of the project. In this paper, we present the architecture of the first iteration of the OpenScienceLink Platform, including detailed information regarding the integration of the background infrastructures and services, as well as the open access data sources utilized.

Portals for Service Oriented Infrastructures

Grid portals enable collaborative environments aiming to provide simple and common Web interfaces to heterogeneous Grid resources and services. However, special factors must be taken into consideration when creating portal applications for business environments. This chapter discusses the approach taken by the Portals technical area of the BEinGRID project, which resulted in the implementation of four software components that address security, user management, file management and management of computational jobs through Grid portals. The components, which were integrated in the Vine Toolkit framework—a collection of Java libraries and User Interfaces for developing Grid applications, are characterised by innovative features that aim to promote the overall business processes and comprise an important improvement towards the business adoption of the Grid. The chapter discusses in detail the technical and business aspects of the components and presents examples of their usage in commercial environments.

An Advanced Query and Result Rewriting Mechanism for Information Retrieval Purposes from RDF Datasources

The volume and variety of data published on the Semantic Web is constantly increasing with a growing number of entities and stakeholders expressing their data in the form of OWL and/or RDFS ontologies. However, a large amount of data is still maintained in relational databases. The recent developments in SPARQL endpoints, such as D2R server, constitute an important step towards the introduction of relational databases in the Semantic Web. However, the underlying models are tightly linked with the data structure and controlled terminologies employed, and hence, they pose a serious barrier to accessing the data by using different languages. In our previous work, we presented an Ontology Alignment Tool for bridging the gap among the terms of two ontologies based on the instantiation of one or more Ontology Patterns. In this paper, we analytically describe a novel approach and an accordingly designed system for enabling users to access data residing in relational databases by using different models and vocabularies than the ones supported by the SPARQL endpoint. The approach is based on the specification and consumption of correspondences with particular focus on SPARQL query and RDF data rewriting mechanisms, which are responsible for making the necessary changes in the queries and optionally results retrieved from the SPARQL endpoint taking into account the models and vocabularies used in each side.

Semantically-Enabled Context-Aware Abbreviations Expansion in the Clinical Domain

Abbreviations (aka acronym, shorthand, initialism) constitute an important part of the textual description of a clinical study. However, detecting their meaning may be rather difficult not only for software agents but also humans since a considerable amount of them is not accompanied by their long form (aka expansion), while, in general, every abbreviation has more than one possible meaning (aka sense) on average. Moreover, even in cases that the abbreviations are specified in the document, detecting their expansion may be still difficult due to partial or non-similarity among them. For this purpose a system has been developed for detecting the meaning of abbreviations used in clinical studies. In the background innovative algorithms and techniques have been used for both abbreviation recognition and suggestion purposes. The evaluation of the system developed was based on an abbreviations-annotated corpus of clinical studies developed in close collaboration with clinical experts.

Bringing it all Together

In this chapter we first summarise the business challenges the innovation opportunities in each thematic area (Sect. 9.1). Then we explain the dependences between the common technical requirements in each area (Sect. 9.2). Afterwards we summarise the common capabilities developed by the BEinGRID programme in order to address these opportunities (Sect. 9.3). Section 9.4 presents examples of scenarios where a large number of these innovations are brought together in order to solve a complex problem. The focus of this section is to stress the “plug-n-play” approach allowed by the BEinGRID Common Capabilities, validating it through integration scenarios, that demonstrates how identified capabilities are combined.