Read about how knowledge management can be made smarter using a knowledge graph built with semantic technology.
Although a lot of effort, time and expenses go into the drug development process (from discovery to market), we all know that no drug is completely free from undesirable effects. As every drug carries the potential for harm, drug safety can only be estimated in terms of the balance between benefits and adverse effects.
Cases of adverse drug reactions (ADRs) experienced by patients could be detrimental for a new product, creating public doubt about the safety of drugs. However, pre-marketing clinical trials often fail to detect ADRs with low incidence rate or when they are caused by drug-drug interaction (which is estimated to be the reason for merely 30% of the total ADRs). This is determined by the fact that these studies do not involve large and diverse patient groups and sometimes don’t last long enough to allow the detection of all potential safety issues.
Here, pharmacovigilance plays a significant role in closely observing ADRs. The European Medicinal Agency’s pharmacovigilance guideline recommends that the Marketing Authorization Holders (MAH) should:
It goes without saying that the Pharma business requires a more systematic and comprehensive approach to monitoring adverse events – especially when it comes to early detection, quality validation and adequate action upon new adverse events. Nowadays, Pharma companies are obliged to maintain a ‘signal detection’ process to identify new adverse events reported in diverse data sources and ensure timely quality assessment and notification of the regulatory bodies according to the severity of the ADR.
The signal detection process established decades ago is no longer capable of handling the exponential growth of safety information that needs to be scanned as well as the complexity of the causal relations between different entities involved in a case report.
We believe that semantic Text Analysis and Knowledge Graphs are enablers for defining an integrated drug safety signals detection process that will lower the time and effort to identify and validate adverse events in order to build a comprehensive and up-to-date safety Knowledge Base.
To this end, Ontotext provides a highly configurable and flexible solution for scientific literature monitoring of adverse events. Our technology goes beyond the standard deep indexing of content, as the solution is able not only to identify potential adverse event mentioned in scientific literature, but also to capture causal relations between key structured ICSR data elements (patient profile (gender, age), drug information (drug name, active ingredients, structured dosage information, co-treatment drug, route of administration), outcome), etc.
The system can process any content feed, which is delivered in a standard format and is currently loaded with an up-to-date subset of case reports from NCBI PubMed. The extracted data from the scientific literature is complemented with structured ADR knowledge from a public Adverse Event Reporting system such as US FAERS and EudraVigilance. The information in it is normalized to any of the established medical and drug terminologies (MedDRA, SNOMED CT, ICD9 & 10, ICPC, MeSH, Disease Ontology, Symptoms Ontology, NCI Thesaurus, FDA NDC, FDA Orange Book, Drugs@FDA, RxNorm, DrugBank, ChEMBL, ChEBI).
Our scientific literature monitoring solution for drug safety signaling utilizes the publicly available safety information in the form of Knowledge Graphs and fuses the knowledge locked in any proprietary safety data with it. In this way, one can trace the safety information throughout the entire drug development lifecycle in a single intelligent system and can automatically detect new safety signals as new data is ingested in the Knowledge Graphs.