基于DWT-SARIMA-LSTM的流感预测模型研究  

Research on Influenza Forecasting Model based on DWT-SARIMA-LSTM

作　　者：胡兆辉 陈兆学[1] HU Zhaohui;CHEN Zhaoxue(School of Health Science and Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学健康科学与工程学院,上海200093

出　　处：《软件工程》2024年第5期56-61,共6页Software Engineering

基　　金：国家中医药管理局中医药创新团队及人才支持计划项目(ZYYCXTD-D-202208)。

摘　　要：为提高流感预测模型的准确率,针对流感数据的季节性与波动性特点,提出利用离散小波分解(DWT)、季节性自回归综合移动平均模型(SARIMA)和长短期记忆神经网络(LSTM)综合建模,构建DWT-SARIMA-LSTM混合预测模型。首先,将流感数据分解为高频成分与低频成分,对低频成分使用SARIMA模型、高频成分使用LSTM模型分别进行预测;其次,将预测值融合得到最终的预测结果;最后,构建流行控制图预警模型。使用从中国香港卫生署官网获得的中国香港地区2010—2019年的流感数据对模型进行预测和验证,其MAE为0.3427,MAPE为8.0973%,RMSE为0.4632,预警模型的准确率为100%,该模型较于如ARIMA-LSTM等其他混合模型有更高的预测精度。In order to improve the accuracy of influenza forecasting models,considering the seasonality and volatility characteristics of influenza data,this paper proposes a hybrid forecasting model DWT-SARIMA-LSTM by integrating Discrete Wavelet Transform(DWT),Seasonal Autoregressive Integrated Moving Average(SARIMA),and Long Short-Term Memory(LSTM)models.The influenza data is first decomposed into high-frequency and lowfrequency components,and the SARIMA model is applied to the low-frequency component and the LSTM model to the high-frequency component for prediction.The predicted values are then combined to obtain the final forecast results.Finally,a warning model of epidemic control chart is constructed.The model is tested on the influenza data obtained from the China,Hong Kong Department of Health official website for the years 2010 to 2019 in the Hong Kong region of China.The results show that the MAE(Mean Absolute Error)of the model is 0.3427,the MAPE(Mean Absolute Percentage Error)is 8.0973%,the RMSE(Root Mean Squared Error)is 0.4632,and the accuracy for the warning model is 100%.The proposed model demonstrates higher prediction accuracy than other hybrid models such as ARIMA-LSTM.

关 键 词：流感预测 小波分解 季节性自回归综合移动平均模型 长短期记忆神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]