基于季节自回归移动平均模型的新疆细菌性痢疾发病趋势预测  

Seasonal autoregressive moving average model-based prediction of bacteriophage dysentery incidence trends in Xinjiang

作　　者：王婷[1] 贺湘焱[2] WANG Ting;HE Xiangyan(School of Public Health,Xinjiang Medical University,Urumqi,Xinjiang 830054,China;People′s Hospital of Xinjiang Uygur Autonomous Region,Urumqi,Xinjiang 830001,China)

机构地区：[1]新疆医科大学公共卫生学院,新疆维吾尔自治区乌鲁木齐830054 [2]新疆维吾尔自治区人民医院科研教育中心,新疆维吾尔自治区乌鲁木齐830001

出　　处：《公共卫生与预防医学》2023年第5期30-34,共5页Journal of Public Health and Preventive Medicine

摘　　要：目的 分析新疆2005—2018年细菌性痢疾的流行特征,探讨季节自回归移动平均模型预测新疆细菌性痢疾发病规律的可行性和适用性,为预防和控制菌痢的决策工作提供科学依据。方法 采用描述性分析对菌痢流行特征进行分析,利用Python软件构建SARIMA模型并对发病趋势进行预测。结果 2005—2018年新疆菌痢平均年报告发病率为35.71/10万,发病高峰集中在6~10月。各年龄组菌痢发病率的差异有统计学意义(χ^(2)=145 605.90,P<0.001),其中0~5岁和>60岁年龄组患病所占比例较大。所得模型为SARIMA(0,1,2)(0,1,1)12,参数均有统计学意义(P<0.05),对残差序列进行Ljung-Box Q检验(Ljung-Box Q test, LBQ),差异无统计学意义(LBQ=0.68,P=0.41),即残差序列为白噪声。预测值与观测值的相对误差范围为3.29%~75.32%,平均相对误差为11.34%。采用构建的最优SARIMA模型,以2005—2018年菌痢月发病率数据为基础,对2019年的发病趋势进行预测,显示其发病率呈下降的态势。结论 SARIMA(0,1,2)(0,1,1)12模型预测新疆菌痢发病率有较好的精确度,可以用于疾病的中期预测。Objective To analyze the epidemiological characteristics of bacillary dysentery in Xinjiang from 2005-2018,to explore the feasibility and applicability of seasonal autoregressive moving average model to predict the incidence pattern of bacillary dysentery in Xinjiang,and to provide a scientific basis for decision-making in the prevention and control of bacillary dysentery.Methods Descriptive analysis was used to analyze the epidemiological characteristics of bacillary dysentery,and Python software was used to construct a SARIMA model and predict the incidence trend.Results The average annual reported incidence rate of bacillary dysentery in Xinjiang from 2005-2018 was 35.71/100000,with peak incidence concentrated in June-October.The difference in the incidence rate of bacillary dysentery among the age groups was statistically significant(X^(2)=145605.90,P<0.001),with a larger proportion of illnesses in the 0-5 and>60 years age groups.The resulting model was SARIMA(0,1,2)(0,1,1)12 with all parameters statistically significant(P<0.05).The Ljung-Box Q test(LBQ)was performed on the residual series and the difference was not statistically significant(LBQ=0.68,P=0.41),i.e.,the residual series was white noise.The relative errors of the predicted and observed values ranged from^(3).29%to 75.32%,with a mean relative error of 11.34%.The optimal SARIMA model constructed was used to predict the incidence trend from 2019 based on monthly incidence data of bacillary dysentery from 2005-2018,which showed a year-on-year decline in incidence.Conclusion The SARIMA(0,1,2)(0,1,1)12 model has good accuracy in predicting the incidence of bacillary dysentery in Xinjiang and can be used for medium-term prediction of the disease.

关 键 词：季节自回归移动平均模型 菌痢 发病预测 

分 类 号：R181[医药卫生—流行病学]