|
Objective: To understand Ningbo Zhenhai District from 1985 to 2007 of typhoid fever , paratyphoid incidence trends ; sliding average time series models , neural networks , and gray series model comparison autoregressive prediction efficiency of typhoid, paratyphoid . Methods: Using 1985-2007 Ningbo Zhenhai sea CDC infectious disease reporting system to collect infectious diseases typhoid , paratyphoid incidence data observed in the region of typhoid , paratyphoid popular ; using the ARIMA dps7.55 software time series prediction model established by quarterly incidence data 1985-2004 , 2005-2007 typhoid , paratyphoid quarter incidence three prediction models tested , in order to compare the three models , neural networks and gray series model pros and cons. Results :1985 - 2007 , Zhenhai District, typhoid , paratyphoid average incidence rate of 23.42/10 million male average annual incidence rate of 24.68/10 million women 22.08/10 million , was the outbreak of the popular state in 1990 , the incidence rate up to 144.56/10 million . ARIMA time series analysis get fit for 50.15% , and validate the model residual sum of squares for 5154.38 ; using neural network analysis of the goodness-of-fit of 73.12% , and validate the model residual sum of squares 3559.24 ; gray series model analysis get fit for 1.20% , validate the model residual sum of squares for 646.58 . Conclusion: The results of this study show Zhenhai, Ningbo, typhoid fever, paratyphoid fever in the country belong to the high incidence to increase supervision of food hygiene , to promote the use of typhoid , paratyphoid vaccine . Three methods of fitting a comparison with these time series data , the neural network model is more applicable to predict Zhenhai, Ningbo, typhoid , paratyphoid incidence trends ARIMA time series model followed the worst gray series model prediction efficiency .
|