Prediction of influenza-like illness cases in Hunan Province based on the long-short term memory model
Objective To establish a model for predicting the incidence trend in the proportion of outpatient visits for influenza-like illness(ILI)in Hunan Province based on a long-short term memory(LSTM)model on the strength of the TensorFlow deep learning framework,and to provide a scientific basis for improving the level of influenza prevention and control in Hunan Province.Methods We collected the data regarding influenza surveillance in Hunan Province from the 1st week of 2010 to the 52nd week of 2022 from the Chinese Influenza Surveillance Information System.The 2010-2019,2010-2020,2010-2021 and 2010-2022 datasets were used sequentially to construct an LSTM model,and the epidemic trend in ILI％was predicted.Results The mean absolute error(MAE)of the four modeling predictions was 0.067,0.060,0.045 and 0.057 respectively.The root mean square error(RMSE)was 0.104,0.104,0.057 and 0.089 respectively.The mean absolute percentage error(MAPE)was 19.191％,24.222％,13.646％and 18.317％respectively.Conclusion The established LSTM model has good fitting results and prediction effect,which can provide a reference basis for influenza prediction,prevention and control in Hunan Province.
influenza-like illnesslong-short term memory modelprediction
NETL
NSTL
万方数据
