作者: Suzanne Pereira , Catherine Letord , Stéfan J. Darmoni , Elisabeth Serrot
DOI: 10.1007/978-2-8178-0285-5_13
关键词: Natural language processing 、 Semantic mining 、 Artificial intelligence 、 Computer science 、 Terminology 、 Indexer 、 Computerized physician order entry
摘要: Background: We report here the implementation of a medication extraction system which extracts drugs names from French clinical texts, on basis Natural Language Processing algorithms (NLP). Within framework PSIP project, Multi- Terminology Indexer (FMTI) is used to extract drug-related information records for detection Adverse Drug Events (ADEs) that can endanger patients. FMTI needed in absence computerized physician order entry (CPOE), because then discharged summaries are only source about prescribed during hospitalization. Methods: uses appropriate terminologies (TUV, ATC, DCI, NC) with different using stemming and phonemization. performed two evaluations measure performances tool codes free-text documents: one algorithm other phonemization algorithm. Results: The performs better results algorithm: 97% precision 95% recall. Conclusion: For we showed semantic mining help drug- related CPOE.