=Paper=
{{Paper
|id=Vol-2042/paper42
|storemode=property
|title=Towards an Ontology for Sharing Information on Pharmacovigilance Signals
|pdfUrl=https://ceur-ws.org/Vol-2042/paper42.pdf
|volume=Vol-2042
|authors=Pantelis Natsiavas,Magnus Wallenberg,Vassilis Koutkias
|dblpUrl=https://dblp.org/rec/conf/swat4ls/NatsiavasK17
}}
==Towards an Ontology for Sharing Information on Pharmacovigilance Signals
==
https://ceur-ws.org/Vol-2042/paper42.pdf
Towards an Ontology for Sharing Information on
Pharmacovigilance Signals
Pantelis Natsiavas1,2, Magnus Wallberg3 and Vassilis Koutkias1,2
1Lab of Computing, Medical Informatics & Biomedical Imaging Technologies, Department
of Medicine, Aristotle University of Thessaloniki, Thessaloniki, Greece
2Institute of Applied Biosciences, Centre for Research & Technology Hellas, Thermi,
Thessaloniki, Greece
3Uppsala Monitoring Centre, Uppsala, Sweden
Abstract. We present the rationale and the key design decisions for an ontology
currently being built to support linking, sharing and further processing of
pharmacovigilance signals. The main goal of the presented ontology is to
provide a conceptual model describing such information towards FAIR data
principles using the Linked Data paradigm.
Keywords: Pharmacovigilance signals, knowledge engineering, FAIR data.
1 Introduction
Signal detection and management is an important part of pharmacovigilance (PV). A
PV signal is defined as “information that arises from one or multiple sources…,
which suggests a new potentially causal association, or a new aspect of a known
association, between an intervention and an event or set of related events...” [1].
Typically, signal information is disseminated via newsletters released by regulatory
authorities, e.g. PV centres and the World Health Organization, in free-text format.
We currently develop an OWL ontology focusing on the publication of PV signal
information in an extendable and reusable manner based on widely-accepted semantic
models. The main goals of building such an ontology are: (a) automatic interlinking
of data presented in different signal reports, to identify relations between data
currently presented in disparate unstructured datasets (i.e. free-text reports), (b)
provenance tracking of PV signal data to support the verification of the source, and
(c) semantically disambiguating information through reference to widely-accepted
thesauri and semantic models. The ultimate goal is to support the semantic enrichment
of PV signal information following the FAIR data principles [2], in order to enhance
the currently applied PV signal investigation practices, through the reuse and better
exploitation of the respective data.
2 Proposed approach
The key design decisions of the ontology focus on the reuse of existing models to
semantically enhance the original PV signal information and facilitate its verification,
reproducibility and further processing (e.g. through semantic reasoning):
The micropublications semantic model [3] is used to annotate evidence-based PV
signal information, facilitating conclusion reproduction, verification etc.
PROV-O [4] is used to model the provenance information of PV signal data items,
(e.g. patient reports from Spontaneous Reporting Systems), as this can be critical to
verify the quality of data for decision making (e.g. duplicate prevention).
The published PV signal reports can be annotated using the Web Annotation Data
Model (WADM) [5], directly relating the information depicted via Linked Data
formalisms with the original unstructured information in free-text format.
Temporal aspects of PV signal data are expressed using the Time Ontology [6]. As
such data typically refer to trends and evolution through time, their temporal
semantic annotation is important, in order to improve the capabilities of identifying
time-related conclusions (e.g. a dependency with a drug release or an epidemic).
The ontology is built according to the NeoN methodology [7] and defines concepts
typically shared when presenting PV signal information in free-text format, e.g.
spontaneous reports, indications of drug usage, adverse effects, rechallenge or final
outcome, disproportionality analysis data etc. These domain specific concepts are
semantically related with the concepts of the above-mentioned models using either
hierarchy relations (e.g. sub-classing) or via the use of OWL properties. Future work
concerns the validation of the ontology with real-world PV reports and its use as
information representation formalism in unstructured data sources analysis for public
health scenarios as part of the platform presented in [8].
References
1. Council for International Organizations of Medical Sciences (CIOMS): Practical Aspects of
Signal Detection in Pharmacovigilance. CIOMS WG VIII report, Geneva (2010).
2. Wilkinson, M.D., et al.: The FAIR Guiding Principles for scientific data management and
stewardship. Sci. data. 3, 160018 (2016).
3. Clark, T., et al.: Micropublications: a semantic model for claims, evidence, arguments and
annotations in biomedical communications. J. Biomed. Semantics. 5, 28 (2014).
4. Belhajjame, et al.: PROV Model Primer, https://w3.org/TR/prov-primer/.
5. Sanderson, R., et al.: Web Annotation Data Model, http://w3.org/TR/annotation-model/.
6. Cox, S., et al.: Time Ontology in OWL, https://www.w3.org/TR/owl-time/.
7. Suárez-Figueroa, M.C., et al.: The NeOn Methodology for Ontology Engineering. In:
Ontology Engineering in a Networked World. pp.9–34, Springer, Berlin, Heidelberg (2012).
8. Natsiavas, P., et al.: A Public Health Surveillance Platform Exploiting Free-Text Sources
via Natural Language Processing and Linked Data: Application in Adverse Drug Reaction
Signal Detection Using PubMed and Twitter. In: Riaño D. et al. (eds) Knowledge
Representation for Health Care. KR4HC 2016, ProHealth 2016. Lecture Notes in Computer
Science, vol 10096. Springer, Cham.