基于字符串相似度匹配算法挖掘健康体检报告疾病分布规律的实践与探讨

Practice and Discussion on Mining Distribution Rule of Diseases in Health Examination Reports Based on String Matching Algorithm

传统的健康体检主检报告存在大量同义词和上下位词,对同一类型疾病文本描述不一致,给后期疾病检索和数据统计造成困难。本文提出基于字符串相似度匹配的健康体检主检报告疾病分布规律挖掘方法,其步骤包括三个方面:首先根据大量调研设计不同疾病的关键搜索字组合,避免疾病重复搜索或漏检;其次利用字符串匹配函数对原始非结构化数据进行检索、排序和相似度计算;最后对统计结果进行关联分析,得出疾病分布规律的报表数据。以长沙市地区2021年1725名医保公务员的主检文本数据为例得出自动化报表数据。方法的优点在于建立起了传统非结构化文本数据与体检报表数据之间的关联关系,对于提高健康管理效率有积极指导意义。

There are a large number of synonyms and hyponyms in the traditional physical examination reports.The different descriptions of the same type of diseases made it difficult for later disease retrieval and data statistics.In this paper,a method based on string matching for mining the distribution rule of diseases in the main report of health examination was proposed.It included three steps.Firstly,the key search word combinations of different diseases were designed according to a large number of investigations to avoid repeated search or missed detection of diseases;Secondly,the string matching function was used to retrieve,sort and calculate the similarity of original unstructured data;Finally,the statistical results were correlated and analyzed,and the report data of disease distribution law were obtained.Taking the main examination text data of 1725 medical insurance civil servants in Changsha in 2021 as an example,the automatic report data was obtained.The advantage of this method was that it established the correlation between traditional unstructured text data and physical examination report data,which was of positive guiding significance for improving the efficiency of health management.