灰色GM(1,1)模型在医院感染发病率预测中的应用  

The applied research of gray GM(1,1) model in forecasting in hospital infection incidence

作　　者：李红[1] 潘东峰[1] 郭忠琴[2] 王岚[1] 梁沛枫[1] 

机构地区：[1]宁夏自治区人民医院统计室,宁夏银川750021 [2]宁夏医科大学公共卫生学院流行病与卫生统计学教研室,宁夏银川750004

出　　处：《宁夏医学杂志》2013年第3期221-223,共3页Ningxia Medical Journal

基　　金：宁夏自治区卫生厅重点科研项目(2011076)

摘　　要：目的探讨灰色GM(1,1)模型医院感染的监控和预警模式的建立提供辅助信息的应用价值。方法以宁夏某医院的医院感染发病率建立GM(1,1)模型,进行样本内回代评价预测模型的可行性,以2011年1—10月的发病率资料作为模型预测评价样本,检测模型预测值和实际值的拟合程度,以相对误差绝对值平均(MAPE)值评价GM(1,1)模型的预测准确性。结果灰色基本模型、季节趋势模型、季节水平模型的MAPE值,分别为28.72%、20.44%、19.65%,模型具有外推价值。季节水平模型为最优模型,模型表达式为X(k+1)=[130.165 411exp(0.016 21*k)-128.635 411]Ct-s,样本外预测MAPE值为16.58%,模型预测精度良好。结论用GM(1,1)模型对医院感染发病率的预测效果良好,但可尝试利用多变量灰色模型,将各影响因素的作用融入模型,结果更具操作性。Objective To investigate the value of the gray GM （1,1） model and to provide supplementary information for the establishment of hospital infection surveillance and early warning mode. Methods GM （ 1, 1 ） model was established according to the infection incidence in Ningxia between Jan. 2005 and Dec. 2010 and viewed incidence data between Jan. and Oct. 2011 as a evaluate sample of model, tested the similarity between the model predictions and actual values and evaluated the predictive accuracy of the GM （ 1,1 ） model. Results The MAPE values of the gray basic model, the seasonal trend model, the seasonal level of model respectively were 28.72%, 20, 44% and 19.65%, the model had a value of extrapolation. Season level model was the optimal model, the model ex- pression was X （k ＋ 1） = [ 130. 165411exp （0.01621 ＊ k） - 128. 635411 ] CT - s, the sample forecast MAPE value was 16.58%, the model had good predictive accuracy. Conclusion The GM （ 1,1 ） model predicting the effect of hospital infection incidence is good, but multivariate gray model can be used if the role of each factor can drive into the model, the results will be more operational.

关 键 词：灰色GM(1 1)模型 医院感染 预测 

分 类 号：R181.3[医药卫生—流行病学]