通讯作者: 周毅, zhouyi@mail.sysu.edu.cn

DOI：10.12201/bmr.202407.00046

Research on Hemorrhagic Fever with Renal Syndrome Incidence Prediction Based on the SARIMA-LSTM Model

• 摘要：目的/意义 探究前沿技术在肾综合征出血热（HFRS）发病率预测中的应用，梳理、组合多种时序分析方法，并评价模型效果，筛选最佳模型。方法/过程 利用2004-2020年全国HFRS发病率数据，分析基于统计学方法的SARIMA、STL-ARIMA、TBATS模型，基于神经网络的NNAR、LSTM模型，基于三种加权方式的SARIMA-LSTM组合模型的预测效果，利用RMSE、MAE、MAPE综合评价模型效果。结果/结论 SARIMA、LSTM为单一模型中较优模型，RMSE、MAE和MAPE分别为：0.01224、0.00981、18.43%，0.00998、0.00705、14.08%；SARIMA-LSTM组合模型相较单一模型效果均有提升，基于误差倒数法的SARIMA-LSTM组合模型为最优模型，三项评价指标值分别为：0.00940、0.00519、9.32%，筛选出的最佳模型以及组合策略，有望为HFRS发病预警系统模型设计提供技术支持与参考。

  关键词： 肾综合征出血热；传染病监测预警；统计学模型；机器学习；SARIMA-LSTM模型; ; ; ; 

   

  Abstract: Purpose/Significance To investigate the application of cutting-edge technologies in predicting the incidence of Hemorrhagic Fever with Renal Syndrome (HFRS), to compile and integrate various time-series analysis methods, and to evaluate model performance in selecting the optimal model. Method/Process Utilizing national HFRS incidence data from 2004 to 2020, the predictive effectiveness of models based on statistical methods: SARIMA, STL-ARIMA, and TBATS, neural network approaches: NNAR, LSTM, and combined models of SARIMA-LSTM with three different weighting schemes were analyzed. The performance of these models is comprehensively assessed using RMSE, MAE, and MAPE. Result/Conclusion The SARIMA and LSTM models are identified as the superior individual models, with their respective performance metrics—RMSE, MAE, and MAPE—recorded as follows: 0.01224, 0.00981, and 18.43% for SARIMA; 0.00998, 0.00705, and 14.08% for LSTM. The combined SARIMA-LSTM model demonstrates enhanced performance compared to individual models. The SARIMA-LSTM model optimized using the reciprocal of error method is deemed the optimal model, achieving significantly reduced error measures with values of 0.00940 for RMSE, 0.00519 for MAE, and 9.32% for MAPE. The selection of this optimal model and the strategic combination approach bodes well to offer technical support and guidance for the development of an early warning system model tailored to forecasting HFRS outbreaks.

  Key words: hemorrhagic fever with renal syndrome; infectious disease surveillance and early warning; statistical model; machine learning; SARIMA-LSTM model

  提交时间：2024-07-18

  版权声明：作者本人独立拥有该论文的版权，预印本系统仅拥有论文的永久保存权利。任何人未经允许不得重复使用。
• html

• 图表

• 金丽珠, 葛辉, 郭青, 李少琼, 杜雪杰. 基于气象因素和机器学习的流感疫情预警研究. 2021. doi: 10.12201/bmr.202012.00008

  赵坚, 赵玉, 徐小卫, 杨亚洲, 虞莹. 基于大数据与5G的传染病智能监测预警处置系统的设计及应用. 2023. doi: 10.12201/bmr.202305.00009

  杜雪杰, 葛辉. 基于BP神经网络方法构建手足口病预测预警模型设计研究. 2021. doi: 10.12201/bmr.202102.00002

  杨林. 冠状动脉周围脂肪衰减指数联合实验室指标预测急性冠脉综合征的研究. 2024. doi: 10.12201/bmr.202409.00001

  阮旭凌, 刘琦, 郭志恒, 晏峻峰. 基于LDA和XGBoost算法的乳腺癌预测模型构建研究. 2022. doi: 10.12201/bmr.202106.00007

  李晓瑛. 基于文献知识发现的传染病继发疾病挖掘分析研究. 2024. doi: 10.12201/bmr.202408.00057

  杜晓菲, 陈杰, 仇丽霞, 柳雅立. BOPPPS联合案例教学法在传染病教学中的应用. 2024. doi: 10.12201/bmr.202406.00015

  王栋, 田勇泉, 钱招昕, 唐艳. 应对传染病流行事件的发热门诊和定点医院空间可达性分析——以湖南省为例. 2021. doi: 10.12201/bmr.202107.00022

  朱志文, 齐国强, 杨恺熠, 赵永根, 余卓, 俞刚. 基于多中心横向联邦学习的以咳嗽为第一主诉的儿童疾病医疗数据共享模型. 2022. doi: 10.12201/bmr.202207.00004

  冯利, 岳小飞. 9种结合式机器学习算法在基于高维数据的肿瘤早期诊断准确性的比较研究. 2021. doi: 10.12201/bmr.202108.00016

• 序号	提交日期	编号	操作
  1	2024-05-30	bmr.202407.00046V1	下载

• 公开评论  匿名评论  仅发给作者 提交评论