摘要
基于大量实测数据探索2型糖尿病的发病规律,寻求其有效的数据处理方法。将数据挖掘技术引入到2型糖尿病数据处理中得出决策分类树,再同医学认识相对照。利用11400条实测数据,采用C4.5算法得出分类树,经实验患病人群的正确识别率为80.90％,未患病人群的正确识别率为92.05％。给出的决策分类树同目前医学上认识的高危因素趋于一致,同时给出了血糖值等于5.85的临界性数值。数据挖掘方法的引入为2型糖尿病数据处理提供了一种新的方法,为其预警、干预和有效控制提供了一种新的解决方案。
The regular pattern of the type 2 diabetes is explored based on the real data and the effective data processing methods are found. A kind of data mining algorithm is used in processing the type 2 diabetes concrete data, and the decision trees are established, and then the result to the knowledge of physic is compared. Taking the advantage of the 11400 items real data and adopting the C4.5 algorithm to establish the decision trees, the rate of the correct recognition of the healthy person is 80.9% and the rate of the correct recognition of the diabetic patient is 92.05% from the testing results. The decision trees are corresponded with the knowledge of physic, and the critical value of the GLU is given out, which is 5.85. A new method is provided to process the type 2 diabetes, and a new scheme in put forward to predicate, intervene and control the type 2 diabetes.
出处
《计算机工程与设计》
CSCD
2004年第11期1888-1892,共5页
Computer Engineering and Design
基金
国家"十五"攻关基金项目(2001BA702B01)。
