论文部分内容阅读
传染病发病率的有效预测在传染病防治工作中意义重大,其预测理论和方法的研究一直是一个热点。现实中影响传染病发病的因素众多、相互关系复杂,各因素的作用机制通常不能或无法用精确的数学语言来准确描述。本文采用基于时间序列的径向基函数(RBF)神经网络模型对传染病发病率进行预测,以实现传染病发病序列的非线性逼近。在实例分析中,以某市1991-2002年乙型肝炎(乙肝)月发病率数据建模,经过网络的不断学习和训练,得到合适的预测模型后对2003年1-6月的月发病率进行预测。通过与2003年1-6月的实际发病率进行比较分析以验证建模的可靠性,并与传统的时间序列模型预测结果进行比较,结果表明应用RBF神经网络模型对乙肝发病率的短期预测精度更高、效果更好。
Effective prediction of the incidence of infectious diseases is of great significance in the prevention and control of infectious diseases, and the research on its prediction theory and methodology has been a hot spot. In reality, there are many factors that affect the incidence of infectious diseases and their interrelations are complex. The mechanism of action of each factor usually can not or can not be accurately described by accurate mathematical language. In this paper, the RBF neural network model based on time series is used to predict the incidence of infectious diseases in order to achieve the non-linear approximation of the infectious disease sequence. In the case study, the monthly incidence of hepatitis B (hepatitis B) was modeled from 1991 to 2002 in a certain city. After continuous learning and training on the network, a suitable prediction model was obtained, and the monthly incidence rate of January-June 2003 Make a prediction Compared with the actual incidence of January-June 2003 to verify the reliability of modeling, and compared with the traditional time series model predictions, the results show that the application of RBF neural network model of short-term prediction accuracy of hepatitis B incidence Higher, better effect.