Abstract: Objective To compare effects of GM（1,1） gray model, Markov chain model, the gray and Markov chain combined model, and SARIMA model on predicting monthly reported cases of hepatitis A. Methods Using data of monthly reported cases of hepatitis A in Jiangxi Province from 2010 to 2014, GM （1,1） gray model, Markov chain model, combined model of gray and Markov chain, and SARIMA model were fitted respectively. Four models were used to predict the monthly reported cases of hepatitis A from January to December 2015 and compare with actual number of cases. The mean absolute percent error （MAPE）, mean error rate （MER）, mean squared error （MSE） and mean absolute error （MAE） were used to evaluate the model prediction effect. Results A total of 2 939 cases of hepatitis A were reported in Jiangxi Province during this period, and showed a downtrend year by year（r_s= -0.838,P<0.01）.SARIMA（0,1,1）（1,0,0）₁₂ was the optimal SARIMA model; the fitting accuracy of GM（1,1） gray model was qualified. The model predicted MAPE from low to high were the gray Markov chain combined model （23.894%）, SARIMA model （25.529%）, GM （1,1） gray model （28.429%）, and Markov chain model （39.426%） ）.MER from low to high were SARIMA model （21.303%）, gray Markov chain combined model （25.574%）, gray model （30.717%）, and Markov chain model （35.203%）.MSE and MAE from low to high were the SARIMA model （45.293, 4.918）, gray Markov chain combined model （47.122, 5.903）, gray model （67.738, 7.091）, and Markov chain model （85.252, 8.126）. Conclusions The grey Markov chain combined model and SARIMA model have better prediction results, and can be used to predict the number of hepatitis A cases.

Key words: Models, statistical, Hepatitis A, Forecasting