Application Of Time Series Model In Predicting The Healthcare-associated Infection In A Tertiary First-class Hospital

Objective: In this study,the data of nosocomial infection cases of patients discharged from a third class a hospital in Shenyang from 2013 to 2019 were retrospectively investigated to understand the overall trend and characteristics of nosocomial infection in the hospital.Based on the monthly incidence of nosocomial infections from 2013 to 2018,ARIMA model and exponential smoothing model were established to predict 2019 The monthly incidence of hospital infection was verified in,and the optimal time series model of the hospital was compared to Understand the rules and trends of nosocomial infections.Methods: Based on the data of nosocomial infection and discharged patients from a top three hospital in Shenyang from 2013 to 2019,the overall trend of nosocomial infection rate was described with charts,and the dynamic changes of nosocomial infection rate were analyzed from the perspective of year and month.The time series model of expert modeler in spss20.0 is used to fit and predict the incidence of hospital infection.The monthly incidence data of hospital infection in 2019 is taken as the validation sample.The fitting results of the two models are combined to comprehensively evaluate the prediction effect and evaluate the prediction effect of the model.Results: 1.In this study,749801 cases were discharged from hospital from 2013 to 2019,including 6927 cases of nosocomial infection,the nosocomial infection rate was 0.92%,the number of discharged patients showed an upward trend,the number of nosocomial infection cases showed a downward trend,the overall nosocomial infection rate also showed a downward trend,the seven-year nosocomial infection rates were: 3.00%,1.87%,0.69%,0.60%,0.70%,0.77%,0.77%,respectively.2.The peak value of the monthly hospital infection rate in different years of the 3A hospital are: January in 2013,2014,2016,2018 and 2019,November in 2015 and August in 2017;the low peak value of the monthly hospital infection rate in different years are: October in 2013,October in 2019,December in 2014,June in 2015,November in 2016,March in 2017 and may in 2018.The high peak of nosocomial infection rate of discharged patients every year is concentrated in January or so,and there is no obvious low peak.The rate of nosocomial infection does not fluctuate seasonally.This study does not consider that the rate of nosocomial infection has seasonal factors.3.ARIMA(0,1,0)model,RMSE = 0.260,MAPE = 22.500,MAE = 0.184,BIC =-2.630,was obtained by using the time series expert modeler in this hospital.The p value of Ljung box Q statistic is 0.256,which is greater than the significant level of 0.05.The simulation effect of this model is good.The actual value of monthly incidence of hospital infection in January December 2019 is basically consistent with the simulation value.4.In the third grade a hospital,RMSE = 0.273,MAPE = 20.358,MAE = 0.197,BIC =-2.481 were calculated by time series expert modeler.The p value of Ljung box Q statistic is 0.618,which is greater than the significant level of 0.05.The selected model is appropriate.The actual value of monthly incidence of hospital infection from January to December 2019 is basically consistent with the simulation value.Conclusion: 1.The overall hospital infection rate of the third class a hospital from 2013 to 2019 shows a downward trend,and the long-term trend is still uncertain.The existing data do not indicate that it is seasonal.In this model,no time series model including seasonal variation is selected.2.The MAPE values of ARIMA(0,1,0)model and exponential smoothing model are 22.500 and 20.358,respectively,which can be used for extrapolation prediction.3.ARIMA model and exponential smoothing model have good fitting effect on the third grade a hospital,and can be used to predict the hospital infection rate.In this study,a more suitable model for the hospital is the exponential smoothing model.4.combined with the prediction model of hospital infection,the incidence rate and confidence interval of hospital infection can be predicted.If the actual incidence rate exceeds the confidence interval,the reasons for the hospital time can be understood and corresponding control measures should be taken.If the actual incidence rate is below the confidence interval,the infection control section is required to investigate whether there is a missing report of hospital infection.