论文部分内容阅读
目的:基于2005—2020年深圳市淋病疫情监测数据，构建自回归移动平均（ARIMA）模型以预测深圳市淋病报告发病率的时间趋势。方法:采用R3.5.0软件建立ARIMA模型，包括模型识别、参数检验和诊断三个步骤。将时间序列分为训练集和验证集，其中2005年1月—2020年5月作为训练集进行模型的建立，2020年6—11月作为验证集进行模型的评估。对比模型的BIC值选择拟合最优的模型，并以平均绝对百分误差（MAPE）为评价标准。结果:根据训练集得出最优模型为ARIMA（0，1，1）（2，1，1）n 12模型（BIC=370.51），应用模型预测2020年6—11月深圳市淋病发病率，发现有周期性波动以及继续下降的趋势，与真实值的发病率趋势相符。该模型MAPE值为18.35%，2020年6—11月的真实值均在预测值的95%n CI内。n 结论:ARIMA（0，1，1）（2，1，1）n 12模型可很好地拟合周期波动和长期趋势，能够应用于预测深圳市淋病发病趋势。n “,”Objective:To construct the autoregressive integrated moving average (ARIMA) model to predict the trend of gonorrhea incidence using the surveillance data from 2005 to 2020 in Shenzhen.Methods:ARIMA model was constructed by R3.5.0 software through three steps which included identification, estimation and diagnosis. The data on gonorrhea incidence was divided into a training dataset from January 2005 to May 2020 for constructing ARIMA model and a hold-out dataset from June to November 2020 for model evaluation. The BIC values were compared to select the best fitting model, and the mean absolute percentage error (MAPE) was used as the evaluation standard.Results:The optimal ARIMA model from the training dataset was ARIMA (0, 1, 1) (2, 1, 1) n 12 model (BIC=370.51) . By applying the model to the gonorrhea incidence of Shenzhen from June to November 2020, the predicted values showed a decreasing trend and a seasonal fluctuation, which holded a similar pattern with the actual values. This model performed well with low MAPE value (18.35%) and the actual incidence from June to November 2020 ranged within the 95% n CI of predicted values.n Conclusions:The ARIMA (0, 1, 1) (2, 1, 1) n 12 model can well fit the long-term trend and seasonal fluctuation in gonorrhea incidence, and can be applied to predict the incidence trend of gonorrhea in Shenzhen.n