【摘要】 目的:首次将知识发现理论引入到2型糖尿病发病相关因素数据处理中,从大量实测数据中识别出有效的、潜在的、有用的、可理解的发病规律。方法:根据2型糖尿病数据的特点,选用数据挖掘C4.5算法对17072条有效的整群抽样横断面健康调查数据进行决策树分类。结果:通过训练模型给出糖尿病患病与否的决策分类树,该决策树可以直观地给出发病相关因素的不同层次的相对影响,经实验测试结果对于未患病的正确识别率为92.05%,对于患病的正确识别率为80.90%,同时得出了血糖值为5.85的分类临界值。结论:决策分类树结果同目前认识的高危因素趋于一致,说明数据挖掘C4.5算法适用于2型糖尿病的发病相关因素数据分析处理,是2型糖尿病数据处理的一种新方法,其在疾病的宏观控制中有着广阔的应用前景。更多还原

【Abstract】 Aims :The technology of knowledge discovery is used to find potential,useful and understandable rules from the data of risk factors related to type2diabetes.Methods:A group of data with risk factors associated with type2diabetes is collected from17,020persons.According to the characters of the data,the algorithm of data mining C4.5is selected to train and test the algorithm model.Results:The decision tree is established from the trained algorithm model.And from the testing results,the correct recognition rate of diabetic patient was80.90%,the correct recognition rate of healthy person was92.05%?Conclusions:The elements revealed by the decision tree are highly relevant to the well-know risk factors with type2diabetes.The algorithm of the data mining C4.5is adapt ed to process of type2diabetes data.It should be a useful method in processing of such kind of data.更多还原