医院住院成本预测

Prediction of Hospitalization Cost

目的分析和预见各种因素对医院成本形成过程的影响程度,加强经济核算。方法应用SPSS、SAS统计软件包对手术人数、院内感染率、病床使用率、平均住院日和鼻咽癌、肝癌、肺癌、女性乳腺癌、食管癌出院人数等因素,分别用最小二乘法和岭估计法对参数进行估计。结果建立了以人均住院成本为因变量的多重线性回归模型。结论岭估计法能克服最小二乘法的缺点,选取适当的k值,消除了自变量间存在的共线性,回归系数估计趋向稳定。

Objective To analyze and predict the influences of different factors on hospitalization cost, to strengthen economic accounting.Methods Different factors were analyzed with SPSS and SAS software. The least-square estimation and ridge estimation was employed to estimate the parameters. Results Multiple linear regression model was established in which the average per capita hospitalization cost was as a dependent variable. Conclusion Ridge estimation could overcome the least-square method′s shortcomings. Multicollinearity among independent variables could be eliminated by choosing appropriate k value. The estimation of regression coefficients tended to be stable.