摘要
目的构建医院早产时间序列的自回归移动平均模型(ARIMA),预测医院早产变化趋势,为合理配置医疗资源、政策制订提供科学依据。方法收集深圳市妇幼保健院2016年逐日早产例数,运用R语言进行时间序列分析,构建ARIMA预测模型,并对预测效果进行评价。结果2016年深圳市妇幼保健院早产1738例,其最佳预测模型为ARIMA(3,1,1),该模型最小赤池信息量准则为1680.67,模型残差序列Ljung-Box检验=0.16,差异无统计学意义(P=0.689),提示残差为白噪声序列,模型拟合良好。模型预测平均相对误差为9.2%,实际值均在预测值95%可信区间内。结论ARIMA(3,1,1)模型能较好地模拟深圳市妇幼保健院早产变化趋势,具有良好的预测效果。
Objective To build a time series autoregressive integrated moving average(ARIMA)model of premature birth in a hospital,to predict the changing trend of premature birth in this hospital,and to provide a scientific basis for rational allocation of medical resources and policy formulation.Methods The number of daily cases concerning premature birth in Shenzhen Maternal and Child Health Care Hospital in 2016 was collected.R language was used to conduct time series analysis,ARIMA prediction model was constructed,and the prediction effect was evaluated.Results There were 1,738 cases of premature birth occurred in Shenzhen Maternal and Child Health Care Hospital in 2016,and the best prediction model was ARIMA(3,1,1).The Akaike Information Criterion of the ARIMA(3,1,1)was 1,680.67.Ljung-Box statistics value=0.16 was not significantly different(P=0.689),suggesting a white noise sequence of residuals with good model fitting.The average relative error between the predictivevalue and the actual value of ARIMA(3,1,1)was 9.2%,and the actual values were within 95%CI of the predictive values.Conclusions The ARIMA(3,1,1)model could forecast the changing trend of premature birth in Shenzhen Maternal and Child Health Care Hospital with good prediction effect.
作者
樊静洁
刘雪芳
刘世新
林一才
牟敬锋
FAN Jing-jie;LIU Xue-fang;LIU Shi-xin;LIN Yi-cai;MOU Jing-feng(Shenzhen Maternal and Child Health Care Hospital,Shenzhen,Guangdong 518017,China;Nanshan District Center for Disease Control and Prevention,Shenzhen,Guangdong 518054,China)
出处
《实用预防医学》
CAS
2020年第4期429-432,共4页
Practical Preventive Medicine
基金
深圳市卫生计生系统科研项目(编号:201607045)
深圳市卫生计生系统科研项目(编号:201607065)。
关键词
早产
ARIMA模型
预测
premature birth
ARIMA model
prediction