OpenPVSignal Knowledge Graph: An openly available data
                         source for pharmacovigilance signal reports
                         Achilleas Chytas 1 and Pantelis Natsiavas 1
                         1
                          Centre for Research and Technology Hellas| Institute of Applied Biosciences, 6th km Charilaou-Thermi 570 01,
                         Thessaloniki, Greece

                                         Abstract
                                         Pharmacovigilance (PV) Signal Reports (SRs) are the consolidation of numerous Individual
                                         Case Safety Reports (ICSRs) by experts for the early detection of causal relationships between
                                         Drugs and Adverse Drug Reactions using statistical correlations. These reports currently exist
                                         in a format not useable by Information and Communications Technology (ICT) systems.
                                         OpenPVSignal model was an effort to bridge the gap between the SRs and ICTs by converting
                                         them to OWL/RDF format. This paper presents the resulting data from the conversion of 108
                                         SRs using OpenPVSignal as the base data model.

                                         Keywords 1
                                         Semantic Web, Real-World Data, Pharmacovigilance

                         1. Introduction
                             Pharmacovigilance (PV) is an integral part of healthcare (HC) systems that plays a pivotal role in
                         the safety and efficacy of pharmaceutical products. Collecting, curating and analyzing data on adverse
                         drug reactions (ADRs) enables HC professionals (HCPs) and regulatory authorities to promptly
                         pinpoint potential safety hazards. Individual Case Safety Reports (ICSRs) typically submitted by HCPs
                         and/or patients are subsequently examined by domain experts via statistical analysis (e.g. via the use of
                         disproportionality analysis metrics). The Uppsala Monitoring Centre, the WHO reference centre for PV
                         (WHO-UMC) maintains and hosts VigiBase (1), the biggest ICSR database in the world, collecting
                         ICSRs from all over the world. They publish PV Signal Reports (PVSR) in a bimonthly newsletter
                         spotlighting new potential drug-ADR relationships. These reports contain curated and novel
                         information that can be crucial to the domain of Drug Safety (DS) but their unstructured format makes
                         it difficult to be taken advantage of by Information and Communications Technology systems.
                             Natsiavas et al (2) have proposed OpenPVSignal, an ontological data model for PVSRs which is
                         maintained in GitHub. This manuscript presents a Knowledge Graph (KG) built upon OpenPVSignal,
                         using 101 PVSRs published by WHO-UMC for a decade between 2012 and 2019.

                         2. Methodology
                            The WHO-UMC data conversion to OWL/RDF has been an iterative process including multiple
                         steps of data quality and validation while a set of researchers that consisted of 4 knowledge engineers
                         and 2 domain experts (one pharmacologist and one physician) were involved in the process. Only a few
                         signals were excluded because they did not focus on drug-ADR interactions, but rather incorrect drug
                         usage or labelling. Figure 1 depicts a PVSR of the presented KG. The KG is openly available to
                         download in Turtle syntax at a GitHub repository https://github.com/inab-certh/OpenPVSignal along
                         with the methodology used for the data validation, while an openly published in a live RDF triple store
                         exists to support exploration http://snf-893389.vm.okeanos.grnet.gr:7200/login.

                         15th International SWAT4HCLS Conference, February 26-29, 2024, Leiden, The Netherlands
                         EMAIL: achytas@certh.gr (A. 1); nbassili@csd.auth.gr (A. 2); pnatsiavas@certh.gr (A.3)
                         ORCID: 0000-0001-8486-011X (A. 1); 0000-0001-6035-1038 (A. 2); 0000-0002-4061-9815 (A. 3)
                                      ©️ 2024 Copyright for this paper by its authors.
                                      Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
                                      CEUR Workshop Proceedings (CEUR-WS.org)


CEUR
                  ceur-ws.org
Workshop      ISSN 1613-0073
Proceedings
   Figure 1 OpenPVSignal instance for PVSR: Vemurafenib and Renal Failure


3. Discussion
OpenPVSignal is an open data source that consists of PVSRs in a FAIR (Findable-Accessible-
Interoperable-Reusable) compliant format. Having these data available in a FAIR format could promote
their systematic reuse and their integration within ICT systems and research pipelines. Additionally, the
OWL semantics along with the rationale of symbolic AI can enable automatic reasoning upon them,
further enhancing their potential by uncovering latent relationships among their Drug and Disease
elements. Thus, the OpenPVSignal KG could play a prominent role in improving the early detection
but also regarding the identification of underlying mechanisms of newly reported ADRs.

4. References
1. Lindquist M. VigiBase, the WHO Global ICSR Database System: Basic Facts. Drug Information
   Journal. 2008 Sep 1;42(5):409–19.

2. Natsiavas P, Boyce RD, Jaulent MC, Koutkias V. OpenPVSignal: Advancing Information Search,
   Sharing and Reuse on Pharmacovigilance Signals via FAIR Principles and Semantic Web
   Technologies. Frontiers in Pharmacology [Internet]. 2018 [cited 2023 Mar 1];9. Available from:
   https://www.frontiersin.org/articles/10.3389/fphar.2018.00609