宁波市镇海区伤寒、副伤寒流行趋势及预测模型评价

发布时间：2019-03-19 09:15

【摘要】：目的:了解宁波市镇海区从1985-2007年的伤寒、副伤寒发病的流行趋势;同时,比较自回归滑动平均时间序列模型、神经网络和灰色数列模型的对伤寒、副伤寒的预测效率。方法:利用宁波市镇海区疾病预防控制中心传染病报告系统收集的1985-2007年传染病伤寒、副伤寒发病资料,来观测该地区伤寒、副伤寒的流行情况;并用dps7.55软件中的ARIMA时间序列、神经网络和灰色数列模型对1985-2004年按季度发病率数据建立预测模型,用2005-2007年的伤寒、副伤寒季度发病率对三种预测模型进行检验,从而比较三种模型的优劣。结果:1985-2007年期间,镇海区伤寒、副伤寒平均发病率为23.42/10万,男性年平均发病率为24.68/10万,女性为22.08/10万,1990年呈爆发流行状态,发病率高达144.56/10万。用ARIMA时间序列分析得到拟合度为50.15%,验证模型的残差平方和为5154.38;用神经网络分析得到拟合度为73.12%,验证模型的残差平方和为3559.24;用灰色数列模型分析得到拟合度为1.20%,验证模型的残差平方和为646.58。结论:本研究结果显示宁波市镇海区伤寒、副伤寒发病在全国属于高发区,要加大对食品卫生的监管,推广使用伤寒、副伤寒疫苗。同时用这些时间序列数据进行三种方法拟合比较得到,神经网络模型更为适用于预测宁波市镇海区伤寒、副伤寒发病趋势,ARIMA时间序列模型其次,灰色数列模型预测效率最差。
[Abstract]:Aim: to understand the epidemic trend of typhoid fever and paratyphoid fever in Zhenhai District of Ningbo City from 1985 to 2007, and to compare the prediction efficiency of typhoid and paratyphoid fever based on autoregressive moving average time series model, neural network and grey series model. Methods: the epidemic situation of typhoid fever and paratyphoid fever in this area was observed by the infectious disease report system collected from 1985 to 2007 in Zhenhai District Disease Control and Prevention Center of Ningbo City, in order to observe the prevalence of typhoid and paratyphoid fever in this area. The ARIMA time series, neural network and grey series model of dps7.55 software were used to establish the forecast model of quarterly incidence data from 1985 to 2004, and three forecasting models were tested with the incidence of typhoid fever and paratyphoid fever from 2005 to 2007, and the incidence rate of typhoid fever and paratyphoid fever from 2005 to 2007 were tested. Thus, the advantages and disadvantages of the three models are compared. Results: from 1985 to 2007, the average incidence of typhoid fever and paratyphoid fever was 23.42 / 100000 in Zhenhai District, 24.68 / 100000 in male and 22.08 / 100000 in female. The incidence was 144.56 / 100000. The fitting degree was 50.15% by ARIMA time series analysis, and the square sum of residual error was 5154.38. The fitting degree was 73.12% by neural network analysis, and the square sum of residual error was 3559.24 by neural network analysis. The fitting degree is 1.20%, and the square sum of residual error is 646.58. Conclusion: the results of this study showed that the incidence of typhoid and paratyphoid fever in Zhenhai district of Ningbo was high in China. It was necessary to strengthen the supervision of food hygiene and popularize the use of typhoid and paratyphoid vaccine. The neural network model is more suitable for predicting the incidence trend of typhoid and paratyphoid fever in the Zhenhai area of Ningbo City. The ARIMA time series model is the second and the grey series model is the least efficient in predicting the occurrence trend of typhoid and paratyphoid fever.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2008
【分类号】：R181.3;R516.3

【相似文献】