摘要
[目的]利用数据挖掘方法衡量各因素对高血压患者住院费用的影响程度,克服传统分析方法局限,为减轻患者负担提供政策参考。[方法]以某市8所不同级别医院2016年2200例高血压住院患者为样本,将住院费用K-means聚类结果作为目标变量,结合单因素分析结果选择输入变量,构建支持向量机模型,评估各影响因素重要性。[结果]多项式核函数支持向量机模型分类准确率最高(93. 84%),住院天数、医院级别、有无合并症及合并症类别、是否手术、性别、付款方式的重要度大于0. 02,对高血压患者的住院费用起主要影响。[结论]数据挖掘方法能够对住院费用进行有效预测;高血压住院负担的控费关键是加强对药费和检查费的监管、缩短住院天数,根本上应充分发挥基层优势,筑起高效的高血压防治体系。
Objective In order to overcome the limitations of traditional analysis methods,the data mining method was used to measure the influence of various factors on hypertensive hospitalization expenses so as to provide policy reference for alleviating the burden of patients. Methods It took 2,200 hypertensive inpatients from 8 hospitals in a city in 2016 as samples,k-means cluster result of hospitalization expenses as target variable.It selected input variables which combined with single-factor analysis to construct support vector machine(SVM)model and evaluate the importance of each influencing factor. Results The classify accuracy rate of polynomial kernel SVM was the highest(93.84%).The importance of stay length,hospital level,had or not complication and type of complication,surgery,gender and payment method was more than 0.02,which had main influence to hypertensive hospitalization expense. Conclusions Data mining method can predict hospitalization cost effectively.The key to control hypertension hospitalization expenses is that to strengthen the supervision of drug and inspection fees and shorten the length of invalid hospitalization days.It should give full play to the advantage of primary-level medical and health care institutions,construct efficiency hypertension prevention and control system.
作者
李相蓉
韩颖
LI Xiang-rong;HAN Ying(Shanxi Medical University,Taiyuan Shanxi 030600,China)
出处
《卫生软科学》
2019年第5期46-50,共5页
Soft Science of Health
基金
山西省卫生总费用核算课题:经常性卫生费用核算技术服务
医疗保险精准抗风险研究
关键词
高血压
住院费用
K-MEANS
聚类分析
支持向量机
hypertension
hospitalization expenses
K-means
cluster analysis
support vector machine
