基于时间序列模型的新疆0~6岁儿童道路交通伤害住院患者人数的预测分析

Time Series Model-based Prediction of the Number of Hospitalizations from Road Traffic Injuries among Children Aged 0-6 Years in Xinjiang

目的分析新疆0~6岁儿童道路交通伤害变化趋势,利用自回归移动平均(autoregressive integrated moving average,ARIMA)模型及指数平滑模型对2011—2021年道路交通伤害住院人数进行分析,模拟验证两种模型适用性后选择最优模型,为本地区儿童道路交通伤害防控及住院人数的动态分析和短期预测提供一定理论支持。方法采取方便抽样的方法,收集新疆7所公立医院2011年1月1日至2021年12月31日0~6岁道路交通伤害入院儿童信息,使用卡方检验分析不同季节、不同受伤类型下道路交通伤害发生情况,分别建立ARIMA模型和指数平滑模型,对模型拟合效果进行评价,验证模型对未来儿童道路交通伤害住院人数预测的适用性。结果本研究共纳入2511例儿童患者,春夏季发生伤害的人数较多,道路交通伤害发生类型以行人及机动车受伤为主。简单季节性指数平滑模型为最优模型,其中稳定系数R2=0.616,均方根误差(root mean square error,RMSE)=6.177,平均绝对误差百分比(mean absolute percentageerror,MAPE)=39.209,平均绝对误差(mean absolute error,MAE)=4.852,正态化贝叶斯信息准则(Bayesian information criterions,BIC)=3.716,模型白噪声检验(P=0.302),该模型可以对未来0~6岁儿童道路交通伤害住院人数进行预测,且模型对年度数据的预测能力优于月度。结论简单季节性指数平滑模型对数据拟合效果更好,可以被用来进行道路交通伤害住院人数的预测研究,未来仍需探索更精准、更全面的预测方法,需采取防控措施预防儿童道路交通伤害的发生。

Objective To analyze changes in the trend of road traffic injuries among children aged 0〜6 years in Xinjiang,the number of road traffic injuries admitted in 2011-2021 by using the autoregression mobile average(autoregressive integrated moving average,ARIMA)model,and theoretical support for the dynamic analysis and short-term prediction of road traffic injury prevention and control of children in the region.Methods A convenience sampling method was used to collect information on children aged 0~6 years who admitted to seven public hospitals in Xinjiang for road traffic injuries from January 1,2011 to December 31,2021.The occurrence of road traffic injuries in difierent seasons and under different injury types were analyzed using chi-square tests.The ARIMA and exponential smoothing models were used to evaluate the model fitting effect,and to verify the applicability of the models for predicting the number of road traffic injuries in children in the future.Results A total of 2511 pediatric patients were included in this study,with a high number of injuries occurring in spring and summer.The types of road traffic injuries were mainly pedestrian and motor vehicle injuries.The simple seasonal exponential smoothing model was the optimal model,based on the stability coefficient F=0.616,root mean square error(RMSE)=6.177,mean absolute percentage error(MAPE)=39.209,mean absolute error(MAE)=4.852,nonnalized Bayesian mfbrmation criterions(BIC)=3.716,and model white noise test(P=0.302).The model can be useful in predicting the number of road traffic injury hospitalizations for children aged 0 to 6 years in the future,and the model has better predictive power than monthly for annual data.Conclusions The simple seasonal exponential smoothing model explained the data better and can be used to conduct prediction studies on the number of road traffic injury hospitalizations.However,more accurate and comprehensive prediction methods need to be explored in the future,and preventive and control measures need to be taken to prevent the occxnrence of road traffic injuries in children.