基于聚类方法的医疗费用数据挖掘研究

Clustering method is applied to the medical records data analysis research

目的医院病案库内数据正大幅增长,但缺乏数据的集成和分析,更谈不上对医学决策和知识的自动获取,为进一步探索和研究住院病人年龄与平均住院日及平均住院费用之间的联系。方法我们使用K-means聚类算法,对医院病案库信息系统中病人年龄、住院时间、住院医疗费用数据进行挖掘,分析内在联系,寻找规律。结果发现30-60岁的住院时间较短,15天以下,但日平均费用并不低,150元以上。而当住院天数超过30天,日平均费用降低至150元以下。说明医疗保险政策还有潜力可挖。结论可以建立一套完整的、合理的、适用的、医疗治疗体制外的保健、保养政策和机构,尽可能的缩短病人平均住院日,降低医院日平均消耗,加大危险期、急性期病情费用的投入,待病人病情稳定后转入相应的保健、保养机构继续调理。

Objective Hospital medical record data inside the library, is the hospital information priority among priorities. But most hospitals on the medical record repository database processing lack of data integration and analysis, not to mention the medical decision making and the automatic acquisition of knowledge, to further explore and study in hospitalized patients age and average hospitalization days and the average hospitalization expenses between contact. Methods We used the K-means clustering algorithml, the hospital medical record repository information system in the age of the patient, the medical expense of hospitalization time, data mining, analysis of immanent connection, to find the law. Results It was found that 30-60, a shorter hospital stay, but the average cost is not low,below 150, and daily average cost below 150 when patient stay in hospital over 30 days, illustrate the medical insurance policy still has the potential to be excavated. Conclusion The country should establish a set of complete, reasonable, applicable, medical treatment and health care system, maintenance policy and mechanism, so as to shorten the average length of stay in hospital, reduce the hospital daily average consumption, and increased health care maintenance mechanism, investment, increase the dangerous period, investment in the acute phase of the illness, the patients in stable condition after the corresponding health maintenance organizations, to continue to adiust.