应用灰色模型对我国结核病死亡率进行趋势分析及预测

目的 分析近年来全国结核病死亡变化趋势并做出预测，为制定合理有效的结核病防治策略提供参考。方法通过全国疾病监测系统死因监测（DSP）获得的全国2005—2012年结核病死亡率分别为6.58、5.14、4.65、4.15、4.05、3.87、3.39和2.29，2005年调整后的DSP系统，包括全国31个省（自治区、直辖市）（不含港澳台地区）的161个监测点。其中城市监测点64个，农村监测点97个。DSP系统总监测人口8153万，约占全国人口的6%，死因分类参照国际疾病分类标准第10版（ICD-10）。通过世界卫生组织（WHO）全球结核病报告（2012版）获得的全国1990—2010年每隔5年的结核病死亡率分别为19.00、13.00、8.70、5.70和3.80。建立GM（1，1）灰色模型，采用MATLAB 7.0分别处理两种来源的结核病死亡率历史数据，获得全国结核病死亡率趋势模型，经拟合检验确认效果后进行外推，得到未来年份的结核病死亡率预测数据。结果 通过GM（1，1）趋势模型、以两种数据来源的数据为依据进行预测，2015、2020、2025年结核病死亡率分别为2.44/10万、1.62/10万、1.07/10万和2.55/10万、1.71/10 万、1.14/10万。作为对比，通过时间序列模型（指数平滑法）以两种数据来源的数据为依据进行预测，2015、2020、2025年结核病死亡率分别为2.27/10万、1.40/10万、0.86/10万和2.54/10万、1.70/10 万、1.14/10万。结论 利用灰色模型GM（1，1）对我国结核病历史死亡率的拟合效果较好，结果显示在整体外部环境和影响因素不变的情况下，我国结核病死亡率呈稳步下降趋势，有希望完成联合国结核病防治千年发展目标。

Application of grey model in analysis and prediction of tuberculosis mortality of China

Objective This study aims to analyze the trend of mortality of tuberculosis of China in recent years and make predictions, to provide reference data for making rational and effective TB control strategies. Methods The mortality from 2005 to 2012 based on national disease surveillance points are respectively 6.58/100 000, 5.14/100 000, 4.65/100 000, 4.15/100 000, 4.05/100 000, 3.87/100 000, 3.39/100 000, and 2.29/100 000. The national DSP system was adjusted in 2005, which nowadays covers 161 sites in 31 provinces （Hong Kong, Macau, Taiwan not included）, including 64 urban sites, 97 rural sites. About 6% of national population, 81.53 million residents are covered by this system and ICD-10 standard is used as classification of cause of death. The mortality based on WHO global tuberculosis report 2012 are 19.00, 13.00, 8.70, 5.70 and 3.80 in every 100 000 population in every 5 years from 1990 to 2010. GM （1,1） grey models are established based on these data by using MATLAB V7.0, then we could get the prediction formulas and predict future TB mortality data after fitting test. Results Via prediction formulas, based on two data resources, the predicted TB mortality of 2015, 2020, 2025 are 2.44/100 000, 1.62/100 000, 1.07/100 000, and 2.55/100 000, 1.71/100 000, 1.14/100 000. Meanwhile, the corresponding predicted results of mortality are 2.27/100 000, 1.40/100 000, 0.86/100 000, and 2.54/100 000, 1.70/100 000, 1.14/100 000, through time series model. Conclusion The GM （1,1） model has showed good effect in simulating historic TB mortality data, and the predicted results indicate that the TB mortality of China will continue to decline in the condition of unchanged external environment and impact factors, and expected to reach the millennium development goals of U.N. in time.