一种快速特征选择方法在医疗数据挖掘中的应用.pdfVIP

数据资源管理与利用 Data Resources Management and Utilization 一种快速特征选择方法在医疗数据挖掘中的应用 ① ② 许杰 缪茹一 摘 要 提出一种基于不一致率的特征选择及数据分类方法，并给出特征选择和分类方法在Weka中的完整实现；利用Weka 中的实现，对某三甲医院的多个医疗数据集进行实验分析；通过Java调用Weka设计开发了一套在线病情分析系统，医生可 通过Web上传医疗数据集，系统后台自动进行数据预处理、特征选择及分类，最终给出病情分类结果，具有较高的诊断决策 支持作用。 关键词　医疗数据 数据挖掘 特征选择 数据分类 Weka Doi:10.3969/j.issn.1673-7571.2016.04.023 [中图分类号] R319 [文献标识码] A A Quick Feature Selection Method and Its Application in Medical Data Mining / XU Jie, MIAO Ru-yi//China Digital Medicine.-2016 11(4): 64 to 66 Abstract In this paper, a method of feature selection and data classification based on inconsistency rate is proposed, and the full implementation of feature selection and classification method in Weka is given. Then, several multi-dimensional medical datasets from a real-world hospital are employed for experimental studies based on Weka. At last, an online patient s condition diagnosis system is developed using Java language and Weka APIs. When doctors upload the medical datasets, the system can automatically clean and preprocess the original data, then do feature selections and classifications, and finally provide the patient s conditions. The system we developed in this paper is valuable as decision supporting doctors for illness diagnosis. Keywords medical dataset, data mining, feature selection, data classification, Weka Fund project Scientific Research Project of Education Department of Zhejiang Province-Research on the Application of Improvement of Outpatient Service Quality Management with Index Analysis (No. Y201016958); 2011 Provincial Medical General Research Program-Research on the Software Development and Application of the Hospital Database Security Defense System (No. 2011KYA068) Corresponding author The Second Affiliated Hospital of Zhejiang University School of Medicine, Han