The Development of Personalized Writing Assistant for Electronic Discharge Summaries Based on Named Entity Recognition

Named entity recognition (NER) is one of the fundamental tasks in natural language processing, with a high utilization value in the medical domain. The electronic discharge summary is a comprehensive clinical document, with the important legal effect especially in medical disputes, which contains patients´ relevant information during hospitalization. Current main writing mode of electronic discharge summary in China is typing along with copying/pasting or modifying on some existing template files with fixed forms, which inevitably leads to writing inefficiency and transcription errors. In order to solve this problem, this paper intelligently analyses some potential writing style using NER and designs a personalized writing assistant scheme to improve efficiency and reduce errors. The NER model trained by Chinese discharge summaries and rich features set has a good performance. The writing assistant monitors some key words typed in the writing process and timely extracts structural information from electronic medical record database as candidate inputs for the writer.