主成分改进的Logistic回归方法探讨及其应用

The Modified Logistic Regression Model and Its Application

目的探讨使用主成分分析处理Logistic回归中共线性问题的方法及其在医学科研中的应用。方法采用多重线性回归中的共线性诊断方法诊断Logistic回归模型的共线性,使用主成分分析处理Logistic回归中的共线性问题,全部计算采用SAS软件。结果在216例高血压脑出血患者的预后影响因素分析中,使用主成分改进的Lo-gistic回归,各估计系数的标准误均有所减小,提示模型结构较为稳定,其结果的可靠性更高。结论使用主成分改进的Logistic回归进行多重共线性的诊断和处理是有效及可行的。

Objective To explore the method that principal component analysis was used to solve the muhicollinearity in the logistic regression analysis and its application in the medical reseach. Methods The data with multivariable muhicollinearity were diagnosed by the method which was used to diagnose multivariable multicollinearity in the multiple linear regression model, treated using principal component analysis.All calculation was completed by SAS. Results Using the logistic regression was improved by principal component analysis to analyze the prognosis of 216 hypertensive intracerebral hemorrhage patients and all estimate standard error had decreased . It meant that the construction of the model was steady and its result was more reliable. Conclusion The new method is effective and feasible for diagnosis and treatment of muhivariable multcollinearity in the logistic regression model analysis.