Spatial-temporal distribution characteristics and prediction of influenza in Hubei in 2009-2020 / 公共卫生与预防医学

ABSTRACT

Objective To explore the spatial-temporal characteristics of influenza epidemic in Hubei from 2009 to 2020, and make short-term prediction to provide reference for influenza prevention and control strategies. Methods Time series seasonal decomposition model and geographic spatial analysis method were used to analyze spatial-temporal evolution characteristics of influenza prevalence in Hubei during 2009-2020. LSTM neural network model was used to predict the monthly influenza incidence from 2020 to 2023. Results Influenza was mainly prevalent in the end of winter and the beginning of spring (December to March) were periods of high influenza incidence. In recent years, the influenza pandemic has shown an increasing trend. Influenza epidemic was characterized by significant spatial differentiation, with “A-shaped point-axis structure” surrounding counties were more severe . The epidemic center of gravity experienced a spatial evolution process from west to east and from north to south. LSTM neural network model predicted that although the influenza incidence rate from January 2020 to December 2023 is lower than that in 2019, it is still at a high level, and shows a peak epidemic in winter and spring. Conclusion Influenza epidemic in Hubei is characterized by a high epidemic period in late winter and early spring, and the southeast of Hubei is the key epidemic area. It is suggested that publicity and prevention and control should be strengthened according to people, time and place, and key populations and areas should be encouraged to receive influenza vaccines in advance.