结合气象与经济因素应用长短期记忆网络模型预测郑州市手足口病的发病趋势
孙进1, 刘今2, 孙明丽3, 刘宇琦4, 姜玥4, 吴伟51. 中国医科大学 公共卫生学院, 沈阳 110122;
2. 中国医科大学 第二临床学院, 沈阳 110004;
3. 中国医科大学 附属盛京医院妇产科, 沈阳 110004;
4. 中国医科大学 第一临床学院, 沈阳 110001;
5. 中国医科大学 公共卫生学院流行病学教研室, 沈阳 110122
收稿日期:
2022-03-04出版日期:
2023-06-30发布日期:
2023-05-31通讯作者:
吴伟E-mail:wuwei@cmu.edu.cn作者简介:
孙进(2000-),男,本科在读关键词: 手足口病, 长短期记忆网络, 传染病预测
Abstract: Objective Using long short term memory (LSTM) model to predict the incidence of hand-foot-mouth disease (HFMD) in Zhengzhou City,to provide theoretical guidance for prevention of HFMD. Methods Spearman correlation analysis was conducted with SPSS 26.0 to identify influencing factors with high correlation. The LSTM neural network was established using Python 3.9. The six models presented in this paper all used the monthly incidence of HFMD in Zhengzhou City from 2010 to 2018 as the training set,and the incidence data in 2019 as the test set. Results Both meteorological and economic factors improved the prediction accuracy of the model. The mean absolute error and rooted mean squared error of the proposed model for the 2019 test set were 231.92 and 273.54,respectively,both of which were better than other LSTM models and superior to commonly used ARIMA models. Conclusion Adding meteorological and economic factors into LSTM models can improve prediction accuracy and guide prevention and control of HFMD.
Key words: hand-food-mouth disease, long short term memory, infectious disease prediction
PDF全文下载地址:
https://journal.cmu.edu.cn/CN/article/downloadArticleFile.do?attachType=PDF&id=3228