差分自回归移动平均乘积季节模型预测广州市肺结核发病趋势

Application of multiple seasonal ARIMA model for predicting the incidence trend of tuberculosis in Guangzhou City

目的探讨应用差分自回归移动平均(autoregressive intergrated moving average, ARIMA)乘积季节模型预测广州市肺结核月发病数的可行性,为制定防控措施提供参考依据。方法利用2010年1月至2019年6月广州市肺结核月发病数据建立ARIMA模型,并以2019年7―12月数据对模型的预测效果进行验证。结果 2010―2019年广州市共报告肺结核124 311例,总体呈下降趋势。2月发病数最少,3―4月发病数最多。拟合出的最佳模型ARIMA (0, 1, 1)(0, 1, 1)12对广州市2019年7―12月肺结核月发病数预测结果显示实际值和预测值相对误差范围介于0.08%~11.33%,平均相对误差为1.46%。结论 ARIMA (0, 1, 1)(0, 1, 1)12模型可用于广州市肺结核月发病数的短期预测。

Objective To explore the feasibility of applying the multiple seasonal autoregressive intergrated moving average(ARIMA) model to predict the monthly incidence of tuberculosis in Guangzhou, and to provide evidence for developing prevention and control measures. Methods The ARIMA model was established based on the monthly incidence of tuberculosis in Guangzhou from January 2010 to June 2019, and the prediction effect of the model was verified with the data from July to December 2019. Results A total of 124 311 tuberculosis cases were reported during 2010-2019 in Guangzhou, showing an overall decreasing trend, with the lowest incidence in February and the hightest in March to April. Using the best fitted model ARIMA(0, 1, 1)(0, 1, 1)12 to predict the monthly incidence of tuberculosis in Guangzhou from July to December 2019, the results showed that the relative error between the actual value and predicted value ranged from 0.08% to 11.33%, and the average relative error was 1.46%. Conclusion The ARIMA(0, 1, 1)(0, 1, 1)12 model can be used for short-term prediction of the monthly incidence of tuberculosis in Guangzhou.