Objective  To compare the performance of seasonal autoregressive integrated moving average (SARIMA) model, Winter linear and seasonal exponential smoothing model, Census X12 seasonal decomposition model, and linear combination prediction model in incidence prediction of hand, foot and mouth disease (HFMD) for facilitating incidence prediction of the disease.

  Methods  Data on reported monthly HFMD incidence from January 2008 through December 2019 were collected from the dataset published by China Center for Disease Control and Prevention and those on annual population during the same period were collected from China Statistical Yearbook 2020 for calculating the monthly incidence of HFMD in China in the period; then monthly incidence of HFMD in China from January 2008 to December 2018 was used as sample modeling data to construct SARIMA model, Winter linear and seasonal exponential smoothing model, Census X12 seasonal decomposition model and linear combination prediction model; finally the monthly incidence of HFMD in China from January through December 2019 was used as the out-of-sample evaluation prediction data to evaluate prediction efficacy of the four models.

  Results  The mean absolute deviation (MAD), mean square error (MSE), mean absolute percentage error (MAPE) were 10.311, 30.757, 1.725% for SARIMA model, 14.433, 112.847, 2.415% for Winter linear and seasonal exponential smoothing model, 8.424, 12.007, 1.409% for Census X12 seasonal decomposition model, and 9.334, 18.847, 1.562% for linear combination prediction model, respectively. The optimum model established was Census X12 seasonal decomposition model, followed by linear combination prediction model, SARIMA model, and Winter linear and seasonal exponential smoothing model.

  Conclusion  The established Census X12 seasonal decomposition model could well predict the incidence of HFMD in China and the utilization of the model could facilitate developing strategies on HFMD prevention and control.