关联规则在小样本临床资料中的应用

目的:探索关联规则在小样本临床资料中的应用。方法:采用数据挖掘技术中的关联规则挖掘方法,以列联表的确切概率和可信度的提高来进行规则剪除,结合实例并与Logistic回归分析进行比较。结果:应用关联规则分析冠心病发生的危险因素同Logistic回归分析的结果相近,但关联规则分析可以提供更多的信息。结论:使用Fisher确切概率法作为度量函数对小样本资料进行关联规则分析是可行的,可以弥补Logistic逐步回归应用的缺陷,又可以较全面地揭示资料中蕴涵的信息。

The Application of Association Rules in Clinical Data of Small Sample

Objective To explore the application of association rules in clinical data of small sample. Methods Using the analysis of association rules of data mining, and the pruning of association rules is based on Fisher's exact test and the improvement of the confidence, and comparing with Logistic regression model through clinical data. Results The result of association rules analysis to dangerous factors of the coronary occurrence is close by Logistic regression model, but association rules analysis can provide more information. Conclusion Using Fisher's exact test as the measure of association rules analysis of the small database, on the one hand it could overcome the limitation of Logistic regression, on the other hand it could discover the full information of the data.