Historically, the main knowledge source for pharmacovigilance consists of spontaneous case reports issued by healthcare professionals and patients and these case reports are usually manually reviewed by evaluators in a qualitative way. Pharmacovigilance has for a long time been confined to a role of passive surveillance of adverse drug reactions (ADRs), which means that measures to look for ADRs are limited to the encouragement of health professionals and others to report safety issues in order to alert the health authorities. This role is indispensable but should now be associated to a “pro-active” approach, in order to identify potential problems and safety risks before they emerge into crises. Additionally, under declaration is a major drawback limiting the ability of pharmacovigilance to process the evaluation of case reports and to detect early signals.
In this context, the development of new data analytic methods and the availability of more efficient computing resources allows considering new data sources for pharmacovigilance, such as electronic health records, administrative claims systems, case-mix databases, social media, search log data and the medical literature. New data mining techniques are being implemented in order to detect safety signals in dominant pharmacovigilance data sources as well as in emerging data sources. Notable systematic efforts relevant with the Research Topic have already been conducted in projects such as the Observational Medical Outcomes Partnership (OMOP)
http://omop.org, the Pharmacoepidemiological Research on Outcomes of Therapeutics by a European Consortium (PROTECT)
http://www.imi-protect.eu, the EU-ADR European project
https://bioinformatics.ua.pt/euadr/Welcome.jsp, the U.S. Food and Drug Administration’s Mini-Sentinel program
https://www.sentinelinitiative.org, and the European Web-RADR project
https://web-radr.eu.
The objective of this Research Topic is to describe the state of advancement of innovative research that relies heavily on data processing for analyzing all possible (traditional and emerging) sources of knowledge for pharmacovigilance. This includes but is not limited to:
• Analyzing big data such as patients’ posts extracted from forums on the internet as well as large observational databases for signal detection,
• Innovative systems considering other data sources than case reports, e.g. medical literature,
• Combining multiple healthcare databases,
• Novel computational signal detection methods for spontaneous reporting systems,
• Influence of the MedDRA hierarchy on case retrieval and signal detection,
• Knowledge engineering techniques for modelling and extracting knowledge on ADRs,
• Joint signal detection through integrative analysis of multiple, heterogeneous data sources,
• Computational methods for signal verification / causality assessment,
• Evaluation / comparative studies of computational signal detection methods,
• Novel tools (e.g. mobile apps and applications at the point-of-care) aiming to reinforce/facilitate ADR reporting,
• Extraction of ADR knowledge from clinical notes in the electronic health record,
• Substantiation of drug safety signals, i.e. providing a biological explanation by exploring mechanistic connections that might explain why a drug produces a specific ADR,
• Public resources intended to support computational process in pharmacovigilance,
• Successful case studies concerning the application of computational methods in pharmacovigilance,
• Opinion and review papers in the domain.
