Objective To explore the association of meteorological factors with influenza incidence in Hunan province with time series model analysis and to evaluate predictive effectiveness of the established model for developing influenza prevention and control strategies.
Methods We retrieved the data on reported influenza incidents and daily meteorological parameters in Hunan province and divided the data into three sets: a dataset from January 2008 through December 2016 for establishing a time series model and two datasets of first and second half year of 2017 for assessing effectiveness of the constructed model in influenza incidence prediction. SPSS 20.0 software was used to process the data collected and Spearman correlation was adopted to analyze the correlation between meteorological factors and influenza incidence.
Results (1) Influenza cases were reported throughout the period from 2008 to 2017 and more cases were reported during winter and spring seasons. (2) The monthly number of reported influenza cases was correlated negatively with monthly average temperature (r = – 0.195, P < 0.05), sunshine hours (r = – 0.483, P < 0.05), and amount of precipitation (r = – 0.116, P < 0.05) but positively with monthly average air pressure (r = 0.195, P < 0.05), relative humidity (r = 0.260, P < 0.05), and maximum wind speed (r = 0.279, P < 0.05). (3) Aafter introducing meteorological factors, the average relative errors of autoregressive integrated moving average (ARIMA) (1, 0, 0) (0, 1, 1) 12 model and ARIMAX model were 43.90% and 31.48% for the prediction of influenza incidence from January to June 2017 and 70.65% and 67.17% from July to December 2017.
Conclusion The predictive effectiveness of ARIMAX model is better than that of ARIMA model after the introduction of meteorological parameters, but the two established models are not good when used to predict the influenza incidence for the period of July – December 2017, suggesting that more factors should be included for improving the ARIMAX model.
DownLoad: