基于堆积策略的电子病历实体识别

随着各国政府对健康医疗信息系统的投入,电子病历信息挖掘得到越来越多学者的关注。与传统的文本相比,电子病历有其自身的特点．。在2010年i2b2举办的评测中,概念抽取任务最好系统的F值为0．8523,与传统的命名实体识别效果有一定差距。使用了CRF、最大熵两种模型建立了baseline系统并且使用堆积策略综合两者的结果,使得系统的F值达到了91．1％。

Concept Extraction in EMR based on Stacking Method

With governments increasing investments in health information systems, information extraction in EMR（ electronic medical record） has drawn more and more scholars＇ attention. Compared with the traditional text, EMR has its own characteristics. In 2010 i2b2/VA challenge, F value of the best system in concept extraction task reaches 0. 8523. There＇s a wedge between concept extraction in EMR and traditional name entity recognitions. In order to extract relevant concepts in EMR more precisely, this article uses CRF, maximum entropy to establish baseline systems. The integrated classifier predictions with the stacking strategy are very remarkable, making the system F value reached 91.1%.