The Epidemic Feature Of HFRS And Prediction Of Elman Neural Network

Objective:Explain the epidemic feature of HFRS since 2000, analysis the epidemic factors of HFRS,getting to know the distribution of host and the rate of Hantavirus, establishing the Elman neural network model to the number of the HFRS patients, offer the information of prevention and cure in order to control the HFRS better.Methods:Using Excel 2003 and statistical software SPSS 13.0 to analyze the official count data of HFRS and host of Hantavirus from HuNan province since 2000.Predict the number of HFRS patients with the MATLAB software and establish the Elman predictive model.Result:From 2000 to 2008,there are 11036 patients of HFRS in HuNan province, the average of disease incidence is 1.88/10 ten thousand, Epidemic areas remain largely unchanged, the number of patients of male is more than the female, the ages of onset are mostly 30-50 years.From the distribution of occupation, the maximum number is peasants (74.35%),and then worker(6.66%), the distribution of different, kinds of rats between indoor and outdoor exist(χ2=295.778,p=0.000),the rattus norvegicus are dominance rats indoor, and apodemus agrarius are dominance rats outdoor. The density of different kinds of rats between indoor and outdoor are different(χ2=92.513,p=0.000),but the Hantavirus portable rate among different kinds of rats totally are not different (χ2=1.678,p=0.795),suggesting that it is possible that there is cross spread among different kinds of rats, and apodemus agrarius are highest, and then rattus norvegicus, they are possibly the principal host. Using the Elman neural network, the number of HFRS of Prediction is 56 while the real number is 47, the number change to 52 after rectifying.Conclusion:The disease incidence in HuNan province are keeping descend during 4 years, the main patients are male, the age of onset is still prime of one's life, the occupation are most peasants and workers. The brown andersoni is the dominant kind indoor while apodemus agrarius outdoor. The prediction and rectified prediction are mostly get close to the real number. It is feasible to predict the number of patients of HFRS with the Elman neural network.