Automatic Mapping Clinical Notes to Medical Terminologies

Proceedings of the 2006 Australasian Language Technology Workshop (ALTW2006), pages 75–82. Automatic Mapping Clinical Notes to Medical Terminologies Jon Patrick, Yefeng Wang and Peter Budd School of Information Technologies University of Sydney NSW 2006, Australia {jonpat,ywang1,pbud3427}@it.usyd.edu.au Abstract Automatic mapping of key concepts from clinical notes to a terminology is an im- portant task to achieve for extraction of the clinical information locked in clinical notes and patient reports. The present pa- per describes a system that automatically maps free text into a medical reference terminology. The algorithm utilises Natu- ral Language Processing (NLP) tech- niques to enhance a lexical token matcher. In addition, this algorithm is able to identify negative concepts as well as performing term qualification. The al- gorithm has been implemented as a web based service running at a hospital to process real-time data and demonstrated that it worked within acceptable time limits and accuracy limits for them. How- ever broader acceptability of the algo- rithm will require comprehensive evalua- tions. 1 Introduction Medical notes and patient reports provide a wealth of medical information about disease and medication effects. However a substantial amount of clinical data is locked away in non- standardised forms of clinical language which could be usefully mined to gain greater under- standing of patient care and the progression of diseases if standardised. Unlike well written texts, such as scientific papers and formal medical re- ports, which generally conform to conventions of structure and readability, the clinical notes about patients written by a general practitioners, are in a less structured and often minimal grammatical form. As these notes often have little if any for- mal organisation, it is difficult to extract infor- mation systematically. Nowadays there is an in- creased interest in the automated processing of clinical notes by Natural Language Processing (NLP) methods which can exploit the underlying structure inherent in language itself to derive meaningful information (Friedman et al., 1994). In principle, clinical notes could be recorded in a coded form such as SNOMED CT (SNOMED International, 2006) or UMLS (Lindberg et al., 1993), however, in practice notes are written and stored in a free text repre- sentation. It is believed that the encoding of notes will provide better information for docu- ment retrieval and research into clinical practice (Brown and Sönksen, 2000). The use of standard terminologies for clinical data representation is critical. Many clinical information systems en- force standard semantics by mandating struc- tured data entry. Transforming findings, diseases, medication procedures in clinical notes into structured, coded form is essential for clinician research and decision support system. Using concepts in domain specific terminology can en- hance retrieval. Therefore, converting free text in clinical notes to terminology is a fundamental problem in many advanced medical information systems. SNOMED CT is the most comprehensive medical terminology in the world and it has been adopted by the Australia government to encode clinical disease and patient reports. The doctors want a system to develop a standard terminology on SNOMED CT for reporting medical com- plaints …

Content truncated to 500 words