Using Natural Language Processing Techniques to Identify Paediatric Drug Interactions

Drug interactions occur as a result of combining multiple drugs, while administering a drug in the context of pre-existing conditions or together with certain foods or beverages. The detection of drug interactions by physicians and pharmacists is essential to avoiding adverse drug reactions (ADR) among patients that can cause further deterioration of the patients’ health. This task is really time-consuming in Romania given that it requires examining each Summary of Product Characteristics (SmPC), which is a specialised drug-related document. In view of providing pharmacists with a technical solution for identifying the interactions between different paediatric drugs more rapidly and more efficiently, the research conducted in this paper by applying NLP techniques resulted in the design of a dedicated web application, named DetIMP (“Detector of Interactions between Paediatric Drugs”). Through DetIMP usage, the detection of interactions is achieved by a two-step approach: the extraction of medical Named Entities from SmPCs, followed by a binary classification to predict the existence of interactions in a given medical text.