dra ft Do We Need a Terminology for Fetal Medicine ?

Objectives: To assess the need for a standard terminology for Fetal Medicine. Methods: We start by establishing a corpus of fetal medicine articles, in order to support term extraction. We evaluate the coverage of the extracted terms in standard terminologies and assess the quality of these terms by manual review and natural language processing tools. Results: Our corpus includes 2,437 articles from the journal Prenatal Diagnosis (390,000 sentences and 6M words). Using Termine, we extracted 4,114 high-frequency terms (1,330 normalized terms), of which 70% were deemed relevant to fetal medicine. Only 38% of these terms could be mapped to the UMLS (essentially to disorder and procedure concepts). Conclusions: This investigation suggests that there is a significant body of fetal medicine terms. These terms are generally poorly covered in standard terminologies. In order to support interoperability and data exchange in the domain of fetal medicine, these terms would benefit from being integrated into existing clinical terminologies or organized into a specific terminology. Introduction The Fetal Medicine domain involves a wide range of perinatal clinical and non-clinical specialists, including OBGYN, pediatricians, geneticists, perinatologists, biologists, and nurses. A crucial need for interoperability naturally arises from this very interdisciplinary field of medical practice. Standard terminologies and specific requirements to facilitate data sharing in perinatal medicine have already been advocated for by healthcare professionals . In fact, some terminologies already exist for specific aspects of fetal medicine, e.g., for clinical genetics and phenotypes, medical procedures, diseases and rare diseases, drugs. However, no terminology provides appropriate coverage of fetal medicine concepts. Moreover, these terminologies tend to overlap. Our motivation here is to assess the need for a standard terminology for Fetal Medicine (FM). Our goal is not to build such a terminology, but rather to explore the domain and assess its coverage by standard terminologies. More specifically, we built a corpus of text from the medical literature, performed term extraction, and assessed the quality of the extracted terms. We then explored the coverage of these terms in standard terminologies through the Unified Medical Language System (UMLS) 5 Methods Our approach to assessing the need for a terminology of fetal medicine can be summarized as follows. We start by establishing a corpus a fetal medicine articles, in order to support term extraction. We evaluate the coverage of the extracted terms in standard terminologies and assess the quality of these terms by manual review and natural language processing tools. Establishing a corpus of Fetal Medicine articles The journal Prenatal Diagnosis has been one of the leading journals in fetal medicine for the past decade. It has an Impact Factor of 2.514 (JCR 2013), corresponding to the first quartile among the 78 journals in the Obstetrics and Gynecology category. Although it is not an open access journal, the electronic version of its content can be downloaded in PDF format. Moreover, unlike most other clinical journal in this domain, Prenatal Diagnosis is exclusively dedicated to fetal medicine and is therefore a very good source of domain-specific terms. As illustrated in Figure 1, We used a PubMed query (%22Prenatal+diagnosis%22[Journal]) to retrieve all the citations from PD in XML format, using the web service . Ebot provided by the National Center for Biotechnology Information (NCBI). We selected the citations corresponding to the past 10 years (2005-2014), and extracted the metadata of interest (title, abstract and DOI). We acquired the full-text of the articles by downloading the PDF documents associated with the DOI. We extracted the text content from the PDF files using Adobe Acrobat Pro X. 1 http://www.ncbi.nlm.nih.gov/Class/PowerTools/eutils/ebot.cgi Submitted to the 2015 AMIA Annual Symposium draft material -please do not cite