Creating a dictionary database of medical terminology for medical professionals

电子病历系统在没有标准化术语的情况下日益广泛地使用，导致了电子病历的积累。为了对这些记录进行自然语言处理，我们创建了ComeJisyo，这是一本由77,775个术语组成的词典，目前公众可以使用。在最新版本完成后，我们对两家医院提供的电子病历进行了词类和字符类型的调查，使用ComeJisyo将电子病历的文本切分准确率提高到90%以上。ComeJisyo还可以比较多个分析的结果(即荟萃分析)，它还包括一个发音关键字，可以实现更实用的假名汉字和语音转换。

The increasingly widespread use of an electronic health record system without a standardized terminology has caused an accumulation of electronic medical records.In order to perform natural language processing of these records, we created ComeJisyo, a dictionary comprised of 77, 775 terms that is currently available to the public. After the latest version was complete, we investigated word class and character type in the electronic medical records provided by two separate hospitals.Using ComeJisyo improves the accuracy of the text segmentation of electronic medical records to more than 90%. ComeJisyo also makes it possible to compare results of multiple analyses(i. e. meta-analysis) and it includes a pronunciation key that enables more practical kana-kanji and speech conversion.