综合性医院数字化档案信息自动分类方法研究.pdfVIP

《自动化技术与应用》 2023年第42卷第7期 计算机与通信技术 Computer and Communication Technology DOI:10.20033/j. 1003-7241.（2023 ）07-0074-04. 综合性医院数字化档案信息自动分类方法研究 孙 娆 （江苏省连云港市第一人民医院，江苏 连云港 222006） 摘 要:针对数字化医疗档案数目急剧增多，不同类型的档案信息分类变成棘手问题，为此提出基于离散核支持向量机的数字化档案 信息自动分类方法。通过向量空间模型将档案文档扩展为向量空间点，将信息特征矢量与词条矢量间的夹角作为数据库与 医院数字化信息的关联性，降均频与规范化处理特征词条权值评估函数，保留权值高的词条，完成数字化档案信息采集；借助 梯度下降原则获得连续学习规律，运用近似法明确数据间的距离值，得到清洗后的数据大小。实验结果表明，该方法拥有极 佳的分类精度，有效缩短分类所耗时长，为综合性医院数字化档案收集与管理提供借鉴。 关键词:数字化档案分类；信息清洗；无监督学习；离散核支持向量机 中图分类号：TP391 文献标识码:A 文章编号:1003-7241(2023)07-0074-04 Research on Automatic Classification Method of Digital Archives Information in General Hospital SUNRao ( The First Peoples Hospital of Lianyungang City, Lianyungang 222006 China ) Abstract:In view of the rapid increase in the number of digital medical records, the classification of different types of file information be- comes a difficult problem. Therefore, an automatic classification method based on discrete kernel support vector machine is pro- posed. Through the vector space model, archive documents are expanded into vector space points, and the Angle between the in- formation feature vector and the entry vector is taken as the correlation between the database and the hospital digital information. The average frequency is reduced and the feature entry weight evaluation function is processed in a standardized manner, and the entries with high weight are retained to complete the collection of digital archive information. Gradient descent principle is used to obtain the continuous learning rule, the approximate method is used to clarify