Research On Predicting Influenza Incidence

The incidence of influenza has been the focus among people,because this kind of disease is contagious,fastly and widely spread and large social harmfulness.Influenza caused a great disease burden and economic burden to the country and humans.In recent years,people infected with bird flu virus incidents had serious consequences.For these reasons,it is necessary to grasp the variation and the prevalence of influenza.The collection of influenza incidence from national influenza surveillance mechanism seriously lagged behind the development of the disease.The types of influenza virus subtype is too much.So it is difficult to predict the incidence of influenza.This study uses three models to predict the incidence of influenza modeling,namely ARIMA model,unbiased GM(1,1)model and BP neural network.On this basis,it improves the BP neural network algorithm.It fits the relative error from the test samples and finally uses the BP neural network algorithm to predict.As a result,the improved BP neural network algorithm can predict the short-term incidence data better than other models.