组合模型在传染病预测中的应用研究

研究传染病预测问题,由于受到外界因素和人体内部因素共同影响,具有不规则变动和非线性动态特点。传染病发生具有季节性、周期性和非规则等变化特点,单一模型只能预测中部分变化特点,预测误差比较大。为降低传染病预测误差,提出一种组合模型的传染病预测方法。首先分别采用ARIMA和LSSVM模型对传染病发生趋势进行预测,然后将两者预测结果输入到LSSVM重新进行学习,最后得到组合模型的预测结果。采用组合模型对某市乙肝发病率进行仿真。结果表明,组合模型降低了乙肝发病率的预测误差,预测结果更加可靠,为传染病预测工作提供新的技术手段。

Study on Combination Model in Prediction of Infectious Diseases

Because infectious diseases have the characteristics of seasonal,cyclical and irregular changing,single model can only predict part changes,and the prediction error is relatively large.In order to reduce forecasting errors of infectious diseases,we put forward a combined model for prediction of communicable diseases.Firstly,ARIMA and LSSVM models were used for the occurrence trend prediction of infectious diseases,and then the forecasting results were input to the LSSVM for learning.Finally,the combined model prediction results were obtained.The composite model was used to simulate the hepatitis B incidence.The results show that the combined model can reduce the prediction error of hepatitis B incidence,and the forecast results are more reliable.