Document-level adverse drug reaction event extraction on electronic health records in Spanish.

Resumen

We outline an Adverse Drug Reaction (ADRs) extraction system for Electronic Health Records (EHRs) written in Spanish. The goal of the system is to assist experts on pharmacy in making the decision of whether a patient suffers from one or more ADRs. The core of the system is a predictive model inferred from a manually tagged corpus that counts on both semantic and syntactically features. This model is able to extract ADRs from disease-drug pairs in a given EHR. Finally, the ADRs automatically extracted are post-processed using a heuristic to present the information in a compact way. This stage reports the drugs and diseases of the document together with their frequency, and it also links the pairs related as ADRs. In brief, the system not only presents the ADRs in the text but also provides concise information on request by experts in pharmacy (the potential users of the system). Keywords: Event Extraction; Adverse Drug Reactions; Text Mining.