College of Economics and Management, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China; School of Science, Nantong University, Nantong 226019, China.
School of Science, Nantong University, Nantong 226019, China.
Public Health. 2021 Dec;201:108-114. doi: 10.1016/j.puhe.2021.09.025. Epub 2021 Nov 22.
The prediction and early warning of infectious diseases is an important work in the field of public health. This study constructed the grey self-memory system model to predict the incidence trend of infectious diseases affected by many uncertain factors.
The design of this study is a combination of the prediction method and empirical analysis.
By organically coupling the self-memory algorithm with the mean GM(1,1) model, the tuberculosis incidence statistics of China from 2004 to 2018 were selected for prediction analysis. Meanwhile, by comparing with the other traditional prediction methods, three representative accuracy check indexes (MSE, AME, MAPE) were conducting for error analysis.
Owing to the multiple time-points initial fields, which replace the single time-points, the limitation of the traditional grey prediction model, which is sensitive to the initial value, is overcome in the self-memory equation. Consequently, compared with the mean GM model and other statistical methods, the grey self-memory model shows significant forecasting advantages, and its single-step rolling prediction accuracy is superior to other prediction methods. Therefore, the incidence of tuberculosis in China in the next year can be predicted as 55.30 (unit: 1/10).
The grey self-memory system model can closely capture the individual random fluctuation in the whole evolution trend of the uncertain system. It is appropriate for predicting the future incidence trend of infectious diseases and is worth popularizing to other similar public health prediction problems.
传染病的预测和预警是公共卫生领域的一项重要工作。本研究构建了灰色自记忆系统模型，以预测受多种不确定因素影响的传染病发病率趋势。
本研究的设计是预测方法和实证分析的结合。
通过将自记忆算法与均值 GM(1,1)模型有机地结合，对 2004 年至 2018 年中国结核病发病率统计数据进行预测分析。同时，通过与其他传统预测方法进行比较，采用三个代表性的精度检验指标（MSE、AME、MAPE）进行误差分析。
由于自记忆方程采用了多个时间点的初始场来代替传统灰色预测模型中的单一时间点初始场，因此克服了传统灰色预测模型对初始值敏感的局限性。与均值 GM 模型和其他统计方法相比，灰色自记忆模型具有显著的预测优势，其单步滚动预测精度优于其他预测方法。因此，可以预测中国下一年的结核病发病率为 55.30（单位：1/10）。
灰色自记忆系统模型可以紧密捕捉不确定系统整体演化趋势中的个体随机波动。它适用于预测传染病的未来发病率趋势，值得在其他类似的公共卫生预测问题中推广应用。
