应用R软件进行logistic回归模型的交互作用分析

发布时间：2018-07-17 20:27

【摘要】：目的应用R软件进行logistic回归模型的交互作用分析,为探讨交互作用提供依据。方法使用R软件,编写程序实现logistic或Cox回归模型三个评价相加交互作用的指标及其可信区间的计算。结果生物学交互作用的评价应该基于是否有相加交互作用,而流行病学研究中常运用logistic回归模型,并纳入乘积项分析因素间交互作用,其是否有意义仅反映相乘交互作用,并不能反映两因素间相加或生物学交互作用的有无。本文通过实例分析,调用基于R软件编写的interact程序,可以直接计算出logistic或Cox回归模型的三个交互作用评价指标(RERI、AP、SI)及其可信区间;并将结果与运用Andersson编制的Excel计算结果相比较,验证了本程序的科学性和准确性。结论应用R软件编制程序,可实现logistic回归模型因素间交互作用和可信区间的计算,为流行病学研究人员分析生物学交互作用提供依据。
[Abstract]:Objective to analyze the interaction of logistic regression model with R software. Methods R software was used to program the calculation of the additive interaction index and its confidence interval of logistic or Cox regression model. Results the evaluation of biological interaction should be based on whether there is additive interaction. Logistic regression model is often used in epidemiological studies, and the product term is used to analyze the interaction between factors. It does not reflect the existence or absence of additive or biological interactions between the two factors. In this paper, by using the interact program based on R software, we can directly calculate the three index of logistic or Cox regression model and its confidence interval, and compare the result with the result of Andersson. The program is verified to be scientific and accurate. Conclusion the calculation of interaction and confidence interval between factors in logistic regression model can be realized by programming with R software, which can provide basis for epidemiological researchers to analyze biological interaction.
【作者单位】：常州市武进区疾病预防控制中心;常州市第一人民医院;
【基金】：常州市武进区科技发展计划项目(WS201432)
【分类号】：O212.1;R181

【相似文献】