The Application Of Statistical Forecasting Model In The Risk Assessment Of The Hospital Infection

Objective To study the applicating value of the time-series models in the prediction of theincidence of infection in hospitals.Methods Stablish ARIMA model， exponential smoothing model， the gray model accordingto the incidence of nosocomial infection from January2005to December2010in Ningxia， a topthree hospital，to view the incidence of infection from January to October2011as the sample datato evaluate the fitting goodness between the predictions and actual values and the predictiveaccuracy of each model， to determine the optimal forecasting model of the specific incidence ofnosocomial infection.Results The forecast MAPE values of ARIMA model， exponential smoothing models， graymodel are12.50，10.41，19.03， the model forecast evaluation rating is good， the exponentialsmoothing model is the optimal model among the prediction models.Conclusion Using the time-series model to predict the incidence of hospital infections obtainthe satisfactory results，it provide a reference for the monitoring and control of the hospitalinfection. Objective To construct logistic regression prediction model for surgery patients withnosocomial infection, and evaluate the effectiveness of model.Methods collect the data of surgical patients of a hospital in Ningxia in2010, build thehospital infection logistic regression model, the model fitting goodness of fit test was testedand the kappa value and the area under ROC curve was calculated.Results the days of hospitalization, diabetes, coma, chemotherapy, secondary diagnosis≥3, the incision level, endotracheal intubation, tracheotomy tube for surgical patients inhospital risk factors for infection. The Kappa value between the predictive value and themeasured was0.54, the area under the ROC curve was91.8%.Conclusion The logistic regression model prediction have high accuracy for surgicalpatients with hospital infection.