Forecasting On The Epidemic Trend Of HIV/AIDS In Henan

Objectives:The first aim is to describe dynamic changes of the HIV/AIDS epidemic in Henan. The second aim is to explore suitable prediction models for the epidemic trend of HIV/AIDS in Henan and verify its effect. The third aim is to make a medium-term forecast for the future development of the AIDS epidemic, which can provide reference data for setting up the strategy and measures of preventing and controlling AIDS. Methods:At first, annual new reported number and deaths data of HIV/AIDS from 2000 to 2014 in Henan were collected. And dynamic indexes of sequence were computed to describe the changing trend of HIV/AIDS during 15 years. After that, trend extrapolation model, ARIMA model and combined model consist of the former two were used to respectively fit new reported number and deaths of HIV/AIDS from 2000 to 2013 in Henan. In addition, above three models were also used to predict the data of HIV/AIDS in 2014. Fitting precision and prediction effect of each model would be evaluated by the MER and the RER. At last, the best one of three kinds of model on fitting precision and prediction effect was selected to forecast the HIV/AIDS epidemic from 2015 to 2017 in Henan. Results:1 The cumulative reported number of HIV/AIDS from 2000 to 2014 in Henan was up to 65295 persons, with a long-term growth trend. The cumulative deaths of HIV/AIDS from 2000 to 2014 in Henan had reached 18384 people, presenting a first and then decreased phenomenon.2 Trend extrapolation modelS function, among all the models of curve estimation, was the best one to fit the epidemic trend of new reported HIV/AIDS from 2000 to 2013 in Henan. The mean VII error rate for new reported HIV/AIDS during the period of 2000 and 2013 was 19.22%, and the relative error rate for new reported HIV/AIDS in 2014 was 15.77%.Among all the models of curve estimation, Quadratic function could better reflect the epidemic trend of dead HIV/AIDS during the period of 2000 and 2013 in Henan. Fitting the deaths of HIV/AIDS from 2000 to 2013, the mean error rate was 16.59%, and the relative error rate for predicting dead HIV/AIDS in 2014 was 13.41%.3 ARIMA modelAmong all the models of ARIMA, ARIMA(1,2,0) was the optimal model to fit the new reported number of HIV/AIDS epidemic from 2000 to 2013 in Henan. The mean error rate for new reported number of HIV/AIDS during the period of 2002 and 2013 was 22.19%, with the relative error rate of 10.57% for new reported HIV/AIDS in 2014.Among all the models of ARIMA, ARIMA(0,2,1) was the best model to fit the deaths of HIV/AIDS epidemic from 2000 to 2013 in Henan. For the deaths of HIV/AIDS from 2002 to 2013, the mean error rate was 23.24%, with the relative error rate of 11.12% for predicting dead HIV/AIDS in 2014.4 Combined modelCombined model consist of S function and ARMA(1,0) was the suited one to fit new reported number of HIV/AIDS epidemic from 2000 to 2013 in Henan. The mean error rate was 16.16% when combined model fit new reported HIV/AIDS during the period of 2000 and 2013. New reported HIV/AIDS in 2014 was predicted, with the relative error rate of 13.14%.Combined model consist of Quadratic function and ARMA(0,1) was the appropriate one to fit the epidemic trend of dead HIV/AIDS from 2000 to 2013 in Henan. The mean error rate was 13.80% when combined model fit dead HIV/AIDS during the period of 2000 and 2013. Dead HIV/AIDS in 2014 was predicted, with the relative error rate of 6.78%.5 The new reported number of HIV/AIDS from 2015 to2017 in Henan was forecasted by combined model consist of S function and ARMA(1,0), and predicted values were 4208, 4258 and 4302 respectively. The deaths of HIV/AIDS from 2015 to2017 in Henan was forecasted by combined model consist of Quadratic function and ARMA(0,1), and predicted values were 979, 715 and 579 respectively. Conclusions:(1) It can be concluded that three kinds of model, such as trend extrapolation model, ARIMA model and combined model consist of the former two, can all be used to fit the epidemic trend of annual new reported number and deaths of HIV/AIDS during the period of 2000 and 2014 in Henan. Compared with the trend extrapolation model and ARIMA model, combined model has more advantage in fitting accuracy and prediction effect.(2) According to the forecast of HIV/AIDS epidemic from 2015 to 2017 in Henan, the number of new reported HIV/AIDS will continue to increase, but growth has been slow. And the deaths of HIV/AIDS have a little fallen.