基于关联规则的医院感染数据挖掘

Data mining of nosocomial infection based on association rules

目的:探讨关联规则挖掘在医院感染诊断中的应用研究。方法:采用关联规则挖掘算法FP-growth算法,通过设置最小支持度和最小可信度对医院感染病例进行关联分析。结果:从解放军总医院第一附属医院提取400例患者的相关数据,设置最小支持度和最小置信度分别为10%和50%,得到关联规则13条,揭示了年龄、是否有介入操作、抗生素使用、住院时间等因素与医院感染的发生和诊断之间有关联关系。结论:在医院感染病例中挖掘关联规则可以发现医院感染与其产生的可能因素间的规则,这些规则为医院感染的诊断和预防提供重要参考依据。

Objective：To explore the application of association rules in nosocomial infection diagnosis. Methods： Association rules analysis was applied to nosocomial infection data processing through setting minimum support and minimum confidence, using FP growth algorithm in the process. Results： Assigning minimum support and minimum confidence as 10 % and 50 % respectively, and nosocomial infection data from 400 patients was analyzed. Thirteen association rules were discovered. A few correlations among age, with or without interventional operations, use or without use of antibiotic, days of stay in the hospital and diagnosis of nosocomial infection were discovered. Conclusion： Rules of nosocomial infection and possible factors it produced can be discovered by employing association rules in analyzing nosocomial infection cases. And these rules provide important reference for diagnostic criteria and measure, for prevention of nosocomial infection.