Objective   To analyze the epidemiological characteristics of hepatitis B in Wuhan from 2011 to 2021 and evaluate the application of season autoregressive integrated moving average (SARIMA) model and long short - term memory (LSTM) model in predicting the incidence of hepatitis B.

  Methods  The epidemiological characteristics of hepatitis B in Wuhan from 2011 to 2021 were analyzed by descriptive epidemiological method. SARIMA model and LSTM model were established to fit and predict the monthly reported incidence of hepatitis B in Wuhan.

  Results  A total of 55 115 cases of hepatitis B were reported in Wuhan from 2011 to 2021, the average annual incidence rate was 47.37/100 000. There was no obvious seasonal incidence peak. The cases were mainly people aged 30–59 years old (60.90%). The majority of the cases were the jobless or unemployed, the unspecified and farmers. The men to women ratio of the cases was 2.17∶1, the average annual reported incidence rates in men and women were 62.64/100 000 and 30.99/100 000, respectively. The average annual reported incidence rates in central urban area (44.93/100000) was lower than that in outer urban area (50.96/100000). The root-mean-square-error of the SARIMA model and LSTM model were 43.24 and 16.55, respectively, and the mean absolute percentage error was 7.60% and 3.05%, respectively.

  Conclusion  Close attention still needs to be paid to the prevention and control of hepatitis B in adults in Wuhan, and the comprehensive prevention and control of hepatitis B should be strengthened, including hepatitis B surveillance, vaccination and health education, in the elderly and young men. The LSTM neural network model has better fitting and prediction performance compared with the SARIMA model.