Extracting and Structuring Drug Information to Improve e-Prescription and Streamline Medical Treatment

Abstract

Currently, physicians are using the patient electronic health record (EHR) to support their practice. The Romanian healthcare system switched to the electronic prescription starting with 2012. Physicians use the electronic medical record and health card to access patient data whenever available. To improve the medical act, we propose a tool supporting the prescription process, structuring and extracting important information from drug characteristics leaflets (prospectus). The application processes data extracted from around 3.000 medical prospectuses using several Romanian language Web sources. The drug leaflet data is structured on sections: therapeutic action, contraindications, mode of administration, adverse reactions, etc. A stemming algorithm has been applied to each section, extracting the root of the word for an easy search. The result is a text in an *.xml file. After structuring step, the application searches in the structured file the necessary information to prescribe the patient’s medication as closely as possible related to patient state. The application suggests all the drugs matching the patient's disease and are not contraindicated, or enter in conflict with other diseases, treatments or allergies of the patient, and the physician may select the best solution for the given situation.