论文部分内容阅读
目的探讨ARIMA模型在我国成人艾滋病一线治疗人数预测中的应用,阐述建模过程,建立预测模型并预测2011年至2012年治疗人数,分析预测影响因素,为今明两年艾滋病治疗资源配置提供参考。方法数据来自国家免费艾滋病抗病毒药物治疗信息管理系统2003年至2009年成人(>15岁)一线新治疗者按感染途径、基线CD4水平分组的季报人数,通过模型识别、参数估计、检验诊断、模型评价,建立艾滋病一线治疗人数ARIMA模型,预测2010年、2011年治疗人数。结果我国免费艾滋病一线成人治疗累计人数与在治人数预测模型均为ARIMA(1,1,0)(0,1,0)4,预测2010年与2011年底累计人数分别为9.9万、12.1万,在治人数分别为7.3万、8.9万,模型预测误差率分别为0.26%、0.67%。我国免费艾滋病一线成人新治人数预测模型为(1,0,0)(0,1,0)4,预测2010年与2011年一线新治人数分别为2.1万、2.4万,每季度平均新治人数为5.2万,新治人数模型预测误差率为2.39%。我国艾滋病一线新治人数表现逐年上升趋势和季节变化特点,而且受到感染途径、入选治疗标准等因素显著影响。结论 ARIMA模型适用于艾滋病治疗人数早期预测。
Objective To explore the application of ARIMA model in predicting the first-line treatment of adult AIDS in China, and to elaborate on the modeling process, establish the predictive model and predict the number of patients to be treated from 2011 to 2012, and to analyze and predict the influencing factors, so as to provide references for the allocation of AIDS treatment resources in the next two years. Methods Data were collected from the national free HIV / AIDS drug treatment information management system. The number of quarterly reports of new-line first-line therapies by means of infection and baseline CD4 level between 2003 and 2009 was estimated by model identification, parameter estimation, test diagnosis, Model evaluation, establishing ARIMA model of first-line AIDS treatment, and forecasting the number of treatment in 2010 and 2011. Results The ARIMA (1, 1, 0) (0, 1, 0) 4 of ARIMA (0,1,0) 4 for the estimated number of first-line treatment of AIDS-free adults in China and the predicted number of free-of-charge AIDS patients in 2009 were 99,000 and 121,000 respectively, The number of people under governance was 73,000 and 89,000, respectively. The prediction error rates of the models were 0.26% and 0.67% respectively. The prediction model of the number of newly diagnosed adults with free AIDS in China is (1,0,0) (0,1,0) 4. The number of first-line new-line doctors in 2010 and 2011 are estimated to be 21,000 and 24,000, respectively, Million, the number of new governance model prediction error rate of 2.39%. The number of newly diagnosed AIDS patients in our country shows a trend of increasing year by year and seasonal variation, and is significantly affected by factors such as the route of infection and the inclusion of treatment criteria. Conclusion The ARIMA model is suitable for the early prediction of the number of AIDS patients.